- Email: [email protected]
The impact of multiple role occupancy on health-related behaviours in Japan: Differences by gender and age
ARTICLE IN PRESS Public Health (2006) 120, 966–975
www.elsevierhealth.com/journals/pubh
ORIGINAL RESEARCH
The impact of multiple role occupancy on health-related behaviours in Japan: Differences by gender and age Y. Takedaa,, I. Kawachib, Z. Yamagatac, S. Hashimotod, Y. Matsumurae, S. Ogurif, A. Okayamag a
Office for Cancer Control, Health Service Bureau, Ministry of Health, Labour and Welfare, 1-2-2 Kasumigaseki, Chiyoda-ku, Tokyo 100-8916, Japan b Center for Society and Health, Harvard School of Public Health, Boston, USA c Department of Health Sciences, School of Medicine, University of Yamanashi, Yamanashi, Japan d Department of Hygiene, Fujita Health University, School of Medicine, Aichi, Japan e Division of Health Informatics and Education, National Institute of Health and Nutrition, Tokyo, Japan f Department of Hygiene and Preventive Medicine, Iwate Medical University, School of Medicine, Iwate, Japan g Department of Preventive Cardiology, National Cardiovascular Center, Osaka, Japan Received 12 March 2005; received in revised form 1 May 2006; accepted 14 June 2006 Available online 1 September 2006
KEYWORDS Multiple role occupancy; Health-related behaviour; Gender differences; Age differences; Japan
Summary Objectives: We examined gender and age differences in the impact of multiple role occupancy on health-related behaviours and health status among working age Japanese adults. Methods: We analysed the individually linked, nationally representative data of 5693 respondents aged 20–59, who completed the Comprehensive Survey of the Living Conditions of People on Health and Welfare and the National Nutrition Survey, conducted by the Japanese government in 1995. Results: Younger women benefited from multiple roles (less smoking), while younger men demonstrated more high-risk behaviours (more smoking, heavier drinking). By contrast, middle-aged men benefited from multiple roles (less smoking, fewer health problems), while middle-aged women reported lower health maintenance behaviours (less exercise, fewer health check-ups). Conclusions: Japanese society appears to be undergoing a transition in gender roles, as reflected by age and gender differences in the impact of multiple roles on health and health-related behaviours. Middle-aged males benefit from multiple roles (being
Corresponding author. Tel.: +81 3 3595 2185; fax: +81 3 3595 2169.
E-mail address: [email protected] (Y. Takeda). 0033-3506/$ - see front matter & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2006.06.006
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours
967
the primary bread-winner and being married), while middle-aged women do not. This pattern seems to break down for younger Japanese men and women. & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction The rise of dual earner couples in Western, industrialized countries spurred an intense debate concerning the impact of multiple roles on the mental and physical health of women and men. According to the ‘scarcity hypothesis’, the more roles a woman accumulates, the more conflict and strain she will experience, and the greater her chances of experiencing psychological distress and ill health.1 According to the competing ‘expansion hypothesis’, the costs of added roles are more than offset by the rewards that they bring, such as prestige, selfesteem, social support, financial remuneration, and greater control and power within the family.2,3 Accumulated evidence has laid to rest earlier concerns that women’s rising participation in the labour market threatens their well-being.4 Multiple roles have been linked to positive mental and physical health among both genders.5,6 At the same time, multiple roles can be detrimental for some women under some circumstances, and additionally the effects of the same role combinations may be different for men and for women.7,8 Japanese society has lagged behind Western countries in terms of changing gender roles. Despite having the most long-lived women in the world, Japan ranks 38th in the world on the United Nations Development Program’s gender empowerment index.9 Japanese society has steadfastly maintained a strict division of labour between both genders, so that men have been viewed as the primary bread-winner, whilst women’s roles have been focused primarily on the home and family.10 In recent years, however, even the traditional family arrangements have begun to break down in Japan. The decades-long recession—coupled with unprecedented high levels of male unemployment, as well as the erosion of job security and the lifetime employment system—have steadily undermined the traditional gender norms in that society. The traditional role of the husband as a reliable economic provider has been severely shaken and has possibly reduced men’s bargaining power within marital relationships. At the same time, women have become much less tolerant of traditional gender roles.10,11 Another notable societal trend relevant to the changing gender roles in Japan is the rapid aging of
the population. Population aging has increased the demand for long-term care, which in traditional Japanese society falls upon the shoulders of women (or more precisely, the daughters-in-law of aging parents). The revolt of Japanese women against traditional expectations of care-giving has itself been partly blamed for the declining marriage and fertility rates in that country.12 Against this background of changing social realities, few studies in Japan have examined the impact of multiple roles on health. Accordingly, in the present study we sought to examine the differential impacts of multiple role occupancy on health-related behaviours among young and old men and women. Using recent nationally representative surveys, we set out to test the following hypotheses: 1. Multiple role occupancy is associated with increased risks of stress-related coping behaviours, specifically smoking and alcohol consumption, and also with lower prevalence of health-promoting and maintenance behaviours, such as regular exercise and health check-ups. 2. There is an interaction between multiple role occupancy and age group. Specifically, we expect a more pronounced adverse impact of multiple role occupancy on health-related behaviours among younger adults compared to middle-aged adults, due to a combination of: (a) marital conflicts within younger Japanese couples stemming from women’s rising expectations of shared household responsibilities; and (b) recent threats to the traditional role of men as primary bread-winners, accompanied by their gradual loss of authority and bargaining power within households. 3. The association between multiple role occupancy and health-related behaviours is partly mediated by worries among women and men over economic security and future care-giving demands.
Methods Sample The Comprehensive Survey of the Living Conditions of People on Health and Welfare (CSLCPHW) and
ARTICLE IN PRESS 968 the National Nutrition Survey (NNS) have been carried out nationwide among randomly sampled districts in Japan by the government every three years and annually, respectively.13,14 Both surveys include sections on health and socio-economic status (SES) and trained interviewers visited the respondents’ homes. The study population was the respondents of each survey in 1995. CSLCPHW was administered among 5100 randomly sampled districts nationwide. The household response rate was 91.0%. NNS was carried out independently in 300 districts, which were randomly sampled from the same districts in which CSLCPHW was conducted. The estimated response rate to NNS was 94.9%. Permission was granted by the Japanese government to use the 1995 data. We linked data from the two surveys at the individual level and the final number in the linked data set was 13 270, reduced due to simple nonresponse of subjects, or movement during the 5 month interval between the two surveys. The final linkage rate was thus 93.2%.15 In the present report, we restricted the analyses to the sample of working age adults, 20–59 years old (N ¼ 7326).
Definition of multiple roles Social role occupancy in the present study was defined by a linear summation of four different kinds of roles: (1) marital role (married versus unmarried, widowed, separated or divorced); (2) employment role (working in a paid job versus not working in a paid job, including homemaker); (3) parenting role (living with children under age 20 in the household, versus not living with children); and (4) caregiver role (living with parents over age 70, versus not living with parents). Each respondent was assigned a social role occupancy score ranging from 0 (no role; none of the above roles) to 4 (quadruple roles). In addition, we defined nine kinds of role combinations: (1) employment role only, (2) employment and marital role, (3) employment, marital and parenting role, (4) employment, marital, parenting and caregiver role, (5) marital role only, (6) marital and parenting role, (7) marital, parenting and caregiver role, (8) no role and (9) other combinations.
Assessment of health behaviours and health status Survey respondents were asked about their health behaviours. For smoking and drinking, individuals were categorized into one of three groups: never smoker/drinker, current smoker/drinker and former smoker/drinker. We defined those who drank more
Y. Takeda et al. than one gou in Japanese sake/day and more than three times a week as ‘current drinker’. One gou (about 180 ml) of sake is equivalent to about 22 g of alcohol consumption (about two glasses (240 ml) of wine). We defined three categories of physical activity: (1) no daily exercise due to illness or poor health conditions, (2) no daily exercise due to other reason, (3) daily exercise. We set the second category as the reference group. We additionally asked whether the respondent had had a health check-up in the previous year, with two categories, ‘Yes’ or ‘No’. This variable is an indicator of preventive self-care (as opposed to health-seeking following the onset of illness symptoms). In Japanese society, regular health check-ups are widely practiced as a health maintenance strategy and often provided by employers. We also examined one indicator of self-reported health status, i.e. ‘having subjective health problem(s) affecting one’s daily life, nursing care at home, or being hospitalized’.
Assessment of worries We assessed the respondents’ levels of different kinds of daily worries as a potential psychosocial mediator linking the multiple social roles to health behaviours and health status. The worry level of each respondent was assessed by 22 items requiring ‘yes’ or ‘no’ answers. These series of questions covered various kinds of worry, including worries about health, finances, work and care-giving. We performed principal components analysis on the responses with varimax rotation for women and men separately and found three factors in each gender. Among women, these were: (1) future health and finance worry, (2) care-giving and human relations worry and (3) work worry. For the men, the three factors were: (1) work, human relation and financial worry, (2) future health and finance worry and (3) care-giving worry. The first two factors in each gender showed evidence of satisfactory internal consistency reliability (Cronbach’s a coefficients of 0.62 and 0.55 in women, 0.61 and 0.63 in men, respectively). For each type of worry, we created a total score by summing the responses to each item (0 ¼ for ‘no’, 1 ¼ for ‘yes’). Subsequently, we categorized each worry scale into three groups: no worries (score ¼ 0), low worries (score ¼ 1) and moderate–high worries (scoreX2).
Assessment of covariates We had information on household expenditure during the previous month and adjusted expenditure
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours for household size with the equivalence elasticity set to 0.5.16 Ninety-four yen was equivalent to US$1.00 at the 1995 average exchange rate. We also used domiciliary area (the unit is ‘jyo’, equivalent to 1.653 m2) of the household as another socio-economic indicator. Our survey did not include more conventional measures of SES available in the West, e.g. income, because of issues of cultural sensitivity.
Data analysis We conducted multivariable logistic regression analyses to calculate the odds ratios (OR) and 95% confidence intervals (95% CI) for health outcomes, according to the number of social roles (or the role combination) and worry levels of respondents. Separate analyses, stratified by gender, were performed for the four health-related behaviours (smoking, drinking, exercise and health check-ups) as well as the presence of health problems. Three sets of logistic regression models were implemented: model 1, adjusted for age group; model 2, adjusted for age group, SES (quartile of household expenditures and domiciliary area of residence), and whether the respondent reported a health problem; and model 3, added worry scales as additional mediators after adjusting for all the covariates in model 2. In accordance with our second hypothesis, we stratified our analyses according to the respondents’ age group: younger adults (20–39) and middle-aged adults (40–59). In addition to the main effects model described above, we also checked for interactions between age group and social role occupancy. All analyses in the present study were performed using the Statistical Analysis System, Release 8.02 (SAS Institute Inc., NC, USA).
Results Sample distribution (Table 1) After exclusion of households that did not provide complete information on relationships among family members, our final data set comprised 3011 females and 2682 males, aged 20–59. Women reporting ‘no social role’ consisted mainly of single young women in their 20 s, as well as students and other unpaid workers (excluding housewives). The corresponding category among men similarly consisted of single young men in their 20 s and students.
969
Number of social roles and health-related outcomes in women (Table 2) Among women overall, the risk of smoking declined as their number of social roles increased. We also found a significantly higher risk of smoking associated with moderate–high levels of worry about care-giving and human relations. The risk of drinking was nearly doubled among dual role women compared with single role women and the association was more pronounced in younger adults. We found a significantly higher risk of drinking habit associated with lower level of worry about care-giving and human relations. There was a highly significant inverse dose– response gradient between role occupancy and regular exercise, which was primarily observed among middleaged women (P-value for age interaction ¼ 0.07). Multiple social role occupancy was also associated with lower ORs of receiving a recent health check-up, as was having no social role. The OR of receiving a health check-up was also significantly reduced in women who expressed a moderate–high level of worry about care-giving and human relations. Contrary to the hypothesis that the assumption of multiple roles leads to greater role strain, we found no association between increasing role occupancy and self-reports of health problems among women in our sample. In fact, women reporting no roles (compared to single roles) were marginally significantly more likely to report an increased likelihood of reporting health problems (OR ¼ 2.21, 95% CI 0.90–5.44), although the direction of causality in this instance is not clear, i.e. sick women may choose not to take on multiple roles.
Number of social roles and health-related outcomes in men (Table 3) Stratifying the sample according to age group revealed an opposing trend among men: multiple role occupancy was associated with higher ORs of smoking among younger men, but lower ORs in middle-aged men (P-value for age role interactiono0.001). Paradoxically, we found a significantly higher risk of smoking associated with lower levels of worry about work, human relations and present finance, as well as a lower risk of smoking with moderate–high levels of worry about future health and finance. There was a strongly positive dose–response gradient between men’s role occupancy and drinking habit. Interestingly, that gradient appeared stronger among younger men (P-value for interaction between roles and age group ¼ 0.08).
ARTICLE IN PRESS 970
Y. Takeda et al.
Table 1 Distribution of demographic, health behaviours, socio-economic-related variables and social role occupancy Women
Number of respondents Age (years) 20–29 30–39 40–49 50–59 Smoking habit Smoker Ex-smoker Non-smoker Information missing Drinking habit Drinker Ex-drinker Non-drinker Information missing Exercise habit Yes No No (due to poor health condition) Information missing Had health check-ups in the previous year Yes No Information missing Have health problems Yes No Information missing Expenses per family member in the previous month o100 thousand yeny 100–200 200–300 4300 Information missing Domiciliary area of household o10 jyoz 10–30 30–50 50–70 470 Information missing Have paid job Yes No Marital status Married Unmarried Widowed Divorced Have dependent children (o20 years) in the household Yes No
Men
No.
(%)
No.
(%)
3011
(100.0)
2682
(100.0)
664 777 855 715
(22.1) (25.8) (28.4) (23.7)
574 704 794 610
(21.4) (26.2) (29.6) (22.7)
298 64 2014 635
(9.9) (2.1) (66.9) (21.1)
968 237 450 1027
(36.1) (8.8) (16.8) (38.3)
209 31 2135 636
(6.9) (1.0) (70.9) (21.1)
937 60 654 1031
(34.9) (2.2) (24.4) (38.4)
480 1772 131 628
(15.9) (58.9) (4.4) (20.9)
382 1222 48 1030
(14.2) (45.6) (1.8) (38.4)
1494 1421 96
(49.6) (47.2) (3.2)
1726 865 91
(64.4) (32.3) (3.4)
196 2734 81
(6.5) (90.8) (2.7)
165 2441 76
(6.2) (91.0) (2.8)
517 1741 465 145 143
(17.2) (57.8) (15.4) (4.8) (4.7)
457 1566 402 126 131
(17.0) (58.4) (15.0) (4.7) (4.9)
59 1255 1141 407 128 21
(2.0) (41.7) (37.9) (13.5) (4.3) (0.7)
96 1095 1009 349 111 22
(3.6) (40.8) (37.6) (13.0) (4.1) (0.8)
1617 1394
(53.7) (46.3)
2510 172
(93.6) (6.4)
2318 542 64 87
(77.0) (18.0) (2.1) (2.9)
1973 651 16 42
(73.6) (24.3) (0.6) (1.6)
1454 1557
(48.3) (51.7)
1307 1375
(48.7) (51.3)
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours
971
Table 1 (continued ) Women
Have dependent parents (470 years) in the household Yes No Number of social roles No roley Single rolez Double roles Triple rolesyy Quadruple roles(( Combination of social roles Employment role Employment and marital role Employment, marital and parenting role Employment, marital, parenting and caregiver role Marital role Marital and parenting role Marital, parenting and caregiver role No role Other combinations
Men
No.
(%)
No.
(%)
486 2525
(16.1) (83.9)
334 2348
(12.5) (87.5)
129 898 1172 615 197
(4.3) (29.8) (38.9) (20.4) (6.5)
118 554 649 1172 189
(4.4) (20.7) (24.2) (43.7) (7.0)
472 373 422 197 410 672 114 129 222
(15.7) (12.4) (14.0) (6.5) (13.6) (22.3) (3.8) (4.3) (7.4)
518 571 1092 189 27 14 3 118 150
(19.3) (21.3) (40.7) (7.1) (1.0) (0.5) (0.1) (4.4) (5.6)
‘Having a health problem’ was defined as ‘having a subjective health problem affecting daily life, having nursing care at home or
being hospitalized’. 94.0 yen was equivalent to US$1.00 at the 1995 average exchange rate. z 1 jyo is Japanese area unit: equivalent to 1.653 m2. y ‘No role’ means having no social role in the household, namely no paid job, not married, no dependent child(ren) nor dependent parent(s). z ‘Single role’ means playing just one social role in her/his household. ‘Double role’ means playing two social roles in her/his household. yy ‘Triple roles’ means playing three social roles in her/his household. (( ‘Quadruple roles’ means playing four social roles in her/his household. y
Men with higher role occupancy were more likely to report recent health check-ups than men with no or single role. Finally, men with no role had double the risk of reporting a health problem compared to the reference group, and middle-aged men were more likely to report a health problem compared with younger men. Increasing number of roles was associated with lower ORs of reporting health problems and this inverse gradient was particularly marked among middle-aged men (P-value for interactiono0.01). There was a significantly higher risk of reporting a health problem associated with worry about work, human relations and present finance, as well as worry about future health and finance.
Discussion Studies of multiple role occupancy in Western countries have failed to confirm earlier fears that women’s increasing participation in the paid labour
force would lead to role overload and increase their burden of mental and physical morbidity.4 Considerable evidence has supported the validity of the role expansion hypothesis, which suggests that the benefits of added roles for women have outweighed the costs.2,3 Fewer studies have been carried out among men, or directly compared men to women. When gender differences in the impact of multiple roles have been examined, the available evidence suggests that, in spite of the rise in dual earner couples, women continue to bear a disproportionate share of household responsibilities, resulting in the phenomenon of the ‘double shift’.17,18 Additionally, studies have demonstrated that the social context of roles matters in addition to the number of roles occupied by men and women.19,20 Few studies have been published on multiple role occupancy and health in Japan. The country, however, presents an interesting case study of a society undergoing transition. In recent decades, it has moved from a society with highly fixed and traditional gender roles, to an increasingly westernized
ARTICLE IN PRESS 972
Y. Takeda et al.
Table 2 Odds ratios and 95% confidence intervals for women’s health behaviours and health problems with the number of social roles
Age Younger adults (20–39 years) Middle-aged adults (40–59 years) Number of social roles in total age groups (20–59 years) No role Single role Double roles Triple roles Quadruple roles Worry level based on Future health and finance worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Care-giving and human relations worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Number of social roles in younger adults (20–39 years) No role Single role Double roles Triple roles Quadruple roles Number of social roles in middle-aged adults (40–59 years) No role Single role Double roles Triple roles Quadruple roles
Smoking
Drinking
Exercise
Health check-ups Health problems
OR
OR
OR
OR
95% CI
95% CI
95% CI
95% CI
OR
95% CI
1.00
1.00
1.00
1.00
1.00
0.58
0.44–0.76 1.27
0.92–1.75 1.57
1.25–1.97 1.80
1.53–2.11 1.81
1.26–2.62
0.90 1.00 0.98 0.67 0.27
0.45–1.81 1.52 1.00 0.72–1.33 1.70 0.46–0.99 1.56 0.12–0.61 1.24
0.61–3.80 0.88 1.00 1.14–2.53 0.77 0.98–2.49 0.61 0.58–2.64 0.58
0.46–1.67 0.34 1.00 0.60–0.99 0.50 0.44–0.83 0.66 0.36–0.94 1.11
0.22–0.51 2.06 1.00 0.41–0.61 0.82 0.53–0.83 1.07 0.78–1.57 0.50
1.01–4.23
1.00
1.00
1.00
1.00
1.00
0.95
0.63–1.44 0.74
0.45–1.24 0.93
0.65–1.33 1.17
0.90–1.53 7.80
5.40–11.3
1.45
0.85–2.48 0.69
0.33–1.42 1.10
0.68–1.76 1.20
0.84–1.71 5.47
3.33–8.98
1.00
1.00
1.00
1.00
1.00
1.24
0.88–1.77 1.67
1.15–2.43 0.91
0.68–1.23 0.91
0.74–1.13 0.99
0.64–1.52
1.57
1.07–2.29 1.20
0.75–1.91 0.73
0.50–1.07 0.72
0.56–0.94 1.31
0.81–2.12
0.96 1.00 1.16 0.80 0.36
0.44–2.08 1.72 1.00 0.74–1.80 2.48 0.47–1.37 2.13 0.13–0.95 1.19
0.57–5.18 1.12 1.00 1.25–4.93 1.02 0.96–4.74 0.94 0.30–4.63 0.90
0.54–2.33 0.32 1.00 0.66–1.59 0.35 0.56–1.57 0.55 0.42–1.94 1.46
0.20–0.50 2.21 1.00 0.26–0.46 1.02 0.40–0.76 1.30 0.88–2.41 0.35
0.90–5.44
1.24 1.00 0.82 0.55 0.17
0.14–11.0 1.76 1.00 0.53–1.27 1.38 0.31–0.97 1.34 0.04–0.73 1.26
0.21–15.1 0.70 1.00 0.84–2.26 0.65 0.75–2.39 0.47 0.50–3.14 0.44
0.14–3.46 0.11 1.00 0.48–0.89 0.67 0.32–0.71 0.82 0.24–0.82 0.90
0.02–0.53 2.07 1.00 0.51–0.87 0.73 0.60–1.12 0.98 0.56–1.47 0.53
0.55–1.22 0.67–1.70 0.20–1.24
0.49–2.13 0.56–2.99 0.04–2.90
0.47–9.05 0.46–1.17 0.56–1.72 0.19–1.48
Only parts of model 3 results for each outcome are shown.
pattern of active female labour force participation and shared household responsibilities. The present study sought to examine gender and age differences
in the impact of multiple role occupancy on healthrelated outcomes in the context of major societal changes, including the collapse of the lifetime
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours
973
Table 3 Odds ratios and 95% confidence intervals for men’s health behaviours and health problems with the number of social roles
Age Younger adults (20–39 years) Middle-aged adults (40–59 years) Number of social roles in total age groups (20–59 years) No role Single role Double roles Triple roles Quadruple roles Worry level based on Work, human relations and financial worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Future health and finance worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Number of social roles in younger adults (20–39 years) No role Single role Double roles Triple roles Quadruple roles Number of social roles in middle-aged adults (40–59 years) No role Single role Double roles Triple roles Quadruple roles
Smoking
Drinking
Exercise
Health check-ups Health problems
OR
OR
OR
OR
95% CI
95% CI
95% CI
OR
95% CI
1.00
1.00
1.00
1.00
1.00
1.06
0.81–1.39 1.64
1.29–2.08 0.91
0.69–1.20 1.01
0.83–1.23 2.52
1.60–3.96
0.35 1.00 0.98 1.26 1.28
0.19–0.64 0.28 1.00 0.68–1.41 1.68 0.91–1.76 2.08 0.73–2.23 2.08
0.13–0.59 0.87 1.00 1.20–2.34 1.11 1.54–2.80 0.78 1.28–3.36 0.58
0.45–1.69 0.53 1.00 0.77–1.60 1.62 0.56–1.09 1.88 0.31–0.99 1.19
0.35–0.81 2.22 1.00 1.24–2.11 0.57 1.49–2.37 0.30 0.81–1.74 0.37
1.07–4.62
1.00
1.00
1.00
1.00
1.00
1.57
1.13–2.19 1.12
0.84–1.49 0.98
0.72–1.34 1.09
0.87–1.36 1.47
0.93–2.33
0.86
0.61–1.21 0.91
0.67–1.24 0.78
0.54–1.13 1.02
0.79–1.30 1.85
1.15–2.97
1.00
1.00
1.00
1.00
1.00
1.20
0.74–1.93 0.93
0.61–1.41 0.66
0.39–1.11 1.21
0.86–1.72 7.40
4.73–11.6
0.52
0.27–1.00 0.81
0.45–1.47 1.42
0.74–2.71 0.88
0.56–1.40 6.44
3.68–11.3
0.36 1.00 1.28 1.68 4.52
0.19–0.68 0.29 1.00 0.77–2.13 1.39 1.14–2.50 2.33 1.01–20.5 2.91
0.13–0.64 0.99 1.00 0.87–2.21 1.04 1.63–3.33 0.90 1.05–8.10 0.60
0.50–1.98 0.54 1.00 0.61–1.77 2.03 0.60–1.37 1.77 0.17–2.20 0.75
0.35–0.84 2.74 1.00 1.37–3.01 1.20 1.34–2.34 0.61 0.33–1.67 1.41
1.11–6.79
y
0.26 1.00 0.18–0.84 1.72 0.21–1.00 1.75 0.16–1.00 1.80
0.03–2.56
0.51 1.00 0.51–1.82 1.22 0.33–1.18 1.61 0.20–1.06 1.05
0.10–2.58 3.76 1.00 0.76–1.97 0.31 1.01–2.59 0.15 0.59–1.86 0.20
1.00 0.39 0.46 0.40
Models did not converge due to small numbers.
y
1.00 0.93–3.19 0.97 0.95–3.22 0.62 0.86–3.79 0.46
Only parts of model 3 results for each outcome are shown. y
95% CI
0.33–0.98 0.18–0.52 0.16–0.86
0.47–3.05 0.28–1.32 0.17–11.7
0.65–21.6 0.16–0.60 0.07–0.30 0.08–0.52
ARTICLE IN PRESS 974 employment guarantee (and with it, Japanese men’s identity as primary bread-winners), as well as Japanese women’s rising dissatisfaction with traditional gender norms.
Multiple roles and health behaviours among younger (20–39) men and women The increased risks of both smoking and drinking among younger men with multiple roles seem to support our initial hypothesis about the effects of stress associated with juggling the multiple demands of work and family, particularly in the context of rising job insecurity and increasing expectations to share in household responsibilities. However, the finding that multiple role occupancy was associated with decreased risk of smoking among younger women was unexpected, particularly since the prevalence of smoking among young Japanese women has been rising rapidly in recent years.21 We offer two alternative explanations to account for this paradoxical finding. First, it is possible that young Japanese women are beginning to reap the rewards of role expansion, i.e. their increasing participation in the labour market has increased their status and bargaining position within the family and led to a corresponding decline in their stresscoping behaviour. Alternatively, the protective effect of multiple role occupancy on smoking among women may be an expression of traditional control over women’s behaviour. For example, women living with in-laws or children may be more likely to refrain from smoking due to family disapproval. At the same time, however, the higher prevalence of drinking among younger women with multiple roles seems to suggest a stress-coping response. In more detailed analyses (data not shown), the most significant factor associated with drinking among women was being in a parenting role. The gender discrepancy in regular health checkups, i.e. the strong inverse association with multiple role occupancy among younger women, but an association in the opposite direction for younger men, may be explained by differential access to work-based screening programs, rather than by an effect of multiple role occupancy per se (e.g. women not being able to have a check-up due to time pressures).22 In summary, multiple role occupancy among younger Japanese adults was simultaneously associated with both protective and damaging health behaviours. The net effect of these opposing
Y. Takeda et al. trends on future health outcomes among younger Japanese men and women remains to be determined.
Multiple roles and health behaviours among middle-aged (40–59) men and women Middle-aged women with multiple roles were less likely to smoke, but so were middle-aged men. Middle-aged women with multiple roles were no more likely to be drinkers compared to women with fewer roles, whilst we found scant evidence of an association between multiple role occupancy and classical stress-related coping behaviours among middle-aged men. This contrasts with the picture among younger men. It is unlikely that this difference among men reflects any difference in stress at work, since the recession in Japan (and the accompanying rise in job insecurity) affected both younger and older men alike. Instead, this difference may reflect changing expectations of gender roles, specifically for shared household responsibilities between younger women and men. In contrast to younger women, middle-aged women with multiple roles showed a strong inverse gradient in regular exercise, consistent with time constraints to invest in their own health. Overall, the protective effects of multiple roles on smoking appeared to cancel out the adverse effects on exercise, so that there was no association between multiple roles and self-reported health problems in middle-aged women. By contrast, occupying multiple roles appeared to have a net benefit for middleaged men, such that a strong inverse gradient was observed between the number of roles and the ORs of reporting health problems. In summary, we found strong evidence in support of the role expansion hypothesis for middle-aged Japanese men, i.e. the more roles they occupied, the healthier they seemed. Our study was limited in several respects. Firstly, it was cross-sectional in design and hence limited in its ability to offer causal interpretations. Thus, reverse causation cannot be excluded, in which healthy individuals were able to take on more social roles. Secondly, our study examined health-related behaviours, but not health outcomes as ultimate endpoints. Health problem(s) was a crude representation of health status, at best. Thirdly and most importantly, we lacked data on the quality of the different roles. Focusing on role occupancy per se may mask differences in role quality in each situation.20 Nor were we able to test more sophisticated theoretical models, such as
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours family-to-work spillover, or conversely, work-tofamily spillover.23 Notwithstanding these limitations, our study found intriguing gender and age differences in the relationships between multiple roles and health behaviours, within a large, representative sample of a society in transition. The economic and demographic trajectory of Japanese society suggests that the gendered division of labour in traditional households will not be tenable or sustainable in the future. Although our ‘snapshot’ of Japanese adults presents a mixed picture of the influence of multiple roles on the health of women and men, the ultimate balance of health-damaging and health-protecting influences will depend upon how gender relationships evolve within that society.
Acknowledgements This study was supported by Grant No. H11-toukei002 (Principal investigator of the national research group: Prof. Hiroshi Yanagawa, President of Saitama Prefectural University) from the Ministry of Health, Labour and Welfare, Japan. Statement of competing interest: There are no competing interests on this paper.
References 1. Goode WJ. A theory of role strain. Am Sociol Rev 1960;25:483–96. 2. Marks SR. Multiple roles and role strain: some notes on human energy, time and commitment. Am Sociol Rev 1977;42:921–36. 3. Sieber SD. Toward a theory of role accumulation. Am Sociol Rev 1974;39:567–78. 4. Walsh DC, Sorensen G, Leonard L. Gender, health, and cigarette smoking. In: Amick III BC, Levine S, Tarlov AR, Walsh DC, editors. Society and health. New York: Oxford University Press; 1995. p. 131–71. 5. Rodin J, Ickovics JR. Women’s health: review and research agenda as we approach the 21st century. Am Psychol 1990;45:1018–34.
975
6. Thoits PA. Multiple identities and psychological well-being: a reformulation and test of the social isolation hypothesis. Am Sociol Rev 1983;48:174–87. 7. Barnett RC, Baruch GK. Social roles, gender, and psychological distress. In: Barnett RC, Biener L, Baruch GK, editors. Gender and stress. New York: The Free Press; 1987. p. 122–43. 8. Cleary PD, Mechanic D. Sex differences in psychological distress among married people. J Health Soc Behav 1983;24:111–21. 9. United Nations Development Program. Human Development Report 1999: globalization with a human face. New York: Oxford University Press; 1999. 10. National Institute of Population and Social Security Research, Japan. The Second National Survey on Family in Japan; 1998. 11. Cabinet Office, Japan, Gender Equality Bureau. The Public Opinion Survey on a Gender-Equal Society of Japan; 1992, 1997, 2002. 12. Horlacher DE. Aging in Japan: causes and consequences. International Institute for Applied Systems Analysis. Laxenburg, Austria: IIASA; 2002. 13. Ministry of Health, Labour and Welfare, Japan, Statistics and Information Department. The 1995 Comprehensive Survey of the Living Conditions of People on Health and Welfare (CSLCPHW); 1995. 14. Ministry of Health, Labour and Welfare, Japan, Health Service Bureau. The 1995 National Nutrition Survey (NNS); 1995. 15. Hashimoto S, Kawado M, Matsumura Y, et al. Hoken-toukei ni okeru record-linkage no jisshi kanousei (The feasibility study of record-linkage on Japanese health statistics). J Health Welfare Stat 2001;48:1–5 [in Japanese]. 16. Kawachi I, Kennedy BP. The relationship of income inequality to mortality: does the choice of indicator matter? Soc Sci Med 1997;45:1121–7. 17. Hochschild AR. The second shift. New York: Viking; 1989. 18. Bird CE, Fremont AM. Gender, time use and health. J Health Soc Behav 1991;32:114–29. 19. Moen P, Dempster-McClain D, Williams Jr RM. Social integration and longevity: an event history analysis of women’s roles and resilience. Am Soc Rev 1989;54:635–47. 20. Griffin JM, Fuhrer R, Stansfeld SA, et al. The importance of low control at work and home on depression and anxiety: do these effects vary by gender and social class? Soc Sci Med 2002;54:783–98. 21. Honjo K, Kawachi I. Effects of market liberalisation on smoking in Japan. Tob Control 2000;9:193–200. 22. Ministry of Public Management, Home Affairs, Posts and Telecommunications, Japan, Statistics Bureau. The 2002 Labor Force Survey; 2002. 23. Barnett RC, Baruch GK. Women’s involvement in multiple roles and psychological distress. J Pers Soc Psychol 1985;49: 135–45.
www.elsevierhealth.com/journals/pubh
ORIGINAL RESEARCH
The impact of multiple role occupancy on health-related behaviours in Japan: Differences by gender and age Y. Takedaa,, I. Kawachib, Z. Yamagatac, S. Hashimotod, Y. Matsumurae, S. Ogurif, A. Okayamag a
Office for Cancer Control, Health Service Bureau, Ministry of Health, Labour and Welfare, 1-2-2 Kasumigaseki, Chiyoda-ku, Tokyo 100-8916, Japan b Center for Society and Health, Harvard School of Public Health, Boston, USA c Department of Health Sciences, School of Medicine, University of Yamanashi, Yamanashi, Japan d Department of Hygiene, Fujita Health University, School of Medicine, Aichi, Japan e Division of Health Informatics and Education, National Institute of Health and Nutrition, Tokyo, Japan f Department of Hygiene and Preventive Medicine, Iwate Medical University, School of Medicine, Iwate, Japan g Department of Preventive Cardiology, National Cardiovascular Center, Osaka, Japan Received 12 March 2005; received in revised form 1 May 2006; accepted 14 June 2006 Available online 1 September 2006
KEYWORDS Multiple role occupancy; Health-related behaviour; Gender differences; Age differences; Japan
Summary Objectives: We examined gender and age differences in the impact of multiple role occupancy on health-related behaviours and health status among working age Japanese adults. Methods: We analysed the individually linked, nationally representative data of 5693 respondents aged 20–59, who completed the Comprehensive Survey of the Living Conditions of People on Health and Welfare and the National Nutrition Survey, conducted by the Japanese government in 1995. Results: Younger women benefited from multiple roles (less smoking), while younger men demonstrated more high-risk behaviours (more smoking, heavier drinking). By contrast, middle-aged men benefited from multiple roles (less smoking, fewer health problems), while middle-aged women reported lower health maintenance behaviours (less exercise, fewer health check-ups). Conclusions: Japanese society appears to be undergoing a transition in gender roles, as reflected by age and gender differences in the impact of multiple roles on health and health-related behaviours. Middle-aged males benefit from multiple roles (being
Corresponding author. Tel.: +81 3 3595 2185; fax: +81 3 3595 2169.
E-mail address: [email protected] (Y. Takeda). 0033-3506/$ - see front matter & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2006.06.006
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours
967
the primary bread-winner and being married), while middle-aged women do not. This pattern seems to break down for younger Japanese men and women. & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction The rise of dual earner couples in Western, industrialized countries spurred an intense debate concerning the impact of multiple roles on the mental and physical health of women and men. According to the ‘scarcity hypothesis’, the more roles a woman accumulates, the more conflict and strain she will experience, and the greater her chances of experiencing psychological distress and ill health.1 According to the competing ‘expansion hypothesis’, the costs of added roles are more than offset by the rewards that they bring, such as prestige, selfesteem, social support, financial remuneration, and greater control and power within the family.2,3 Accumulated evidence has laid to rest earlier concerns that women’s rising participation in the labour market threatens their well-being.4 Multiple roles have been linked to positive mental and physical health among both genders.5,6 At the same time, multiple roles can be detrimental for some women under some circumstances, and additionally the effects of the same role combinations may be different for men and for women.7,8 Japanese society has lagged behind Western countries in terms of changing gender roles. Despite having the most long-lived women in the world, Japan ranks 38th in the world on the United Nations Development Program’s gender empowerment index.9 Japanese society has steadfastly maintained a strict division of labour between both genders, so that men have been viewed as the primary bread-winner, whilst women’s roles have been focused primarily on the home and family.10 In recent years, however, even the traditional family arrangements have begun to break down in Japan. The decades-long recession—coupled with unprecedented high levels of male unemployment, as well as the erosion of job security and the lifetime employment system—have steadily undermined the traditional gender norms in that society. The traditional role of the husband as a reliable economic provider has been severely shaken and has possibly reduced men’s bargaining power within marital relationships. At the same time, women have become much less tolerant of traditional gender roles.10,11 Another notable societal trend relevant to the changing gender roles in Japan is the rapid aging of
the population. Population aging has increased the demand for long-term care, which in traditional Japanese society falls upon the shoulders of women (or more precisely, the daughters-in-law of aging parents). The revolt of Japanese women against traditional expectations of care-giving has itself been partly blamed for the declining marriage and fertility rates in that country.12 Against this background of changing social realities, few studies in Japan have examined the impact of multiple roles on health. Accordingly, in the present study we sought to examine the differential impacts of multiple role occupancy on health-related behaviours among young and old men and women. Using recent nationally representative surveys, we set out to test the following hypotheses: 1. Multiple role occupancy is associated with increased risks of stress-related coping behaviours, specifically smoking and alcohol consumption, and also with lower prevalence of health-promoting and maintenance behaviours, such as regular exercise and health check-ups. 2. There is an interaction between multiple role occupancy and age group. Specifically, we expect a more pronounced adverse impact of multiple role occupancy on health-related behaviours among younger adults compared to middle-aged adults, due to a combination of: (a) marital conflicts within younger Japanese couples stemming from women’s rising expectations of shared household responsibilities; and (b) recent threats to the traditional role of men as primary bread-winners, accompanied by their gradual loss of authority and bargaining power within households. 3. The association between multiple role occupancy and health-related behaviours is partly mediated by worries among women and men over economic security and future care-giving demands.
Methods Sample The Comprehensive Survey of the Living Conditions of People on Health and Welfare (CSLCPHW) and
ARTICLE IN PRESS 968 the National Nutrition Survey (NNS) have been carried out nationwide among randomly sampled districts in Japan by the government every three years and annually, respectively.13,14 Both surveys include sections on health and socio-economic status (SES) and trained interviewers visited the respondents’ homes. The study population was the respondents of each survey in 1995. CSLCPHW was administered among 5100 randomly sampled districts nationwide. The household response rate was 91.0%. NNS was carried out independently in 300 districts, which were randomly sampled from the same districts in which CSLCPHW was conducted. The estimated response rate to NNS was 94.9%. Permission was granted by the Japanese government to use the 1995 data. We linked data from the two surveys at the individual level and the final number in the linked data set was 13 270, reduced due to simple nonresponse of subjects, or movement during the 5 month interval between the two surveys. The final linkage rate was thus 93.2%.15 In the present report, we restricted the analyses to the sample of working age adults, 20–59 years old (N ¼ 7326).
Definition of multiple roles Social role occupancy in the present study was defined by a linear summation of four different kinds of roles: (1) marital role (married versus unmarried, widowed, separated or divorced); (2) employment role (working in a paid job versus not working in a paid job, including homemaker); (3) parenting role (living with children under age 20 in the household, versus not living with children); and (4) caregiver role (living with parents over age 70, versus not living with parents). Each respondent was assigned a social role occupancy score ranging from 0 (no role; none of the above roles) to 4 (quadruple roles). In addition, we defined nine kinds of role combinations: (1) employment role only, (2) employment and marital role, (3) employment, marital and parenting role, (4) employment, marital, parenting and caregiver role, (5) marital role only, (6) marital and parenting role, (7) marital, parenting and caregiver role, (8) no role and (9) other combinations.
Assessment of health behaviours and health status Survey respondents were asked about their health behaviours. For smoking and drinking, individuals were categorized into one of three groups: never smoker/drinker, current smoker/drinker and former smoker/drinker. We defined those who drank more
Y. Takeda et al. than one gou in Japanese sake/day and more than three times a week as ‘current drinker’. One gou (about 180 ml) of sake is equivalent to about 22 g of alcohol consumption (about two glasses (240 ml) of wine). We defined three categories of physical activity: (1) no daily exercise due to illness or poor health conditions, (2) no daily exercise due to other reason, (3) daily exercise. We set the second category as the reference group. We additionally asked whether the respondent had had a health check-up in the previous year, with two categories, ‘Yes’ or ‘No’. This variable is an indicator of preventive self-care (as opposed to health-seeking following the onset of illness symptoms). In Japanese society, regular health check-ups are widely practiced as a health maintenance strategy and often provided by employers. We also examined one indicator of self-reported health status, i.e. ‘having subjective health problem(s) affecting one’s daily life, nursing care at home, or being hospitalized’.
Assessment of worries We assessed the respondents’ levels of different kinds of daily worries as a potential psychosocial mediator linking the multiple social roles to health behaviours and health status. The worry level of each respondent was assessed by 22 items requiring ‘yes’ or ‘no’ answers. These series of questions covered various kinds of worry, including worries about health, finances, work and care-giving. We performed principal components analysis on the responses with varimax rotation for women and men separately and found three factors in each gender. Among women, these were: (1) future health and finance worry, (2) care-giving and human relations worry and (3) work worry. For the men, the three factors were: (1) work, human relation and financial worry, (2) future health and finance worry and (3) care-giving worry. The first two factors in each gender showed evidence of satisfactory internal consistency reliability (Cronbach’s a coefficients of 0.62 and 0.55 in women, 0.61 and 0.63 in men, respectively). For each type of worry, we created a total score by summing the responses to each item (0 ¼ for ‘no’, 1 ¼ for ‘yes’). Subsequently, we categorized each worry scale into three groups: no worries (score ¼ 0), low worries (score ¼ 1) and moderate–high worries (scoreX2).
Assessment of covariates We had information on household expenditure during the previous month and adjusted expenditure
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours for household size with the equivalence elasticity set to 0.5.16 Ninety-four yen was equivalent to US$1.00 at the 1995 average exchange rate. We also used domiciliary area (the unit is ‘jyo’, equivalent to 1.653 m2) of the household as another socio-economic indicator. Our survey did not include more conventional measures of SES available in the West, e.g. income, because of issues of cultural sensitivity.
Data analysis We conducted multivariable logistic regression analyses to calculate the odds ratios (OR) and 95% confidence intervals (95% CI) for health outcomes, according to the number of social roles (or the role combination) and worry levels of respondents. Separate analyses, stratified by gender, were performed for the four health-related behaviours (smoking, drinking, exercise and health check-ups) as well as the presence of health problems. Three sets of logistic regression models were implemented: model 1, adjusted for age group; model 2, adjusted for age group, SES (quartile of household expenditures and domiciliary area of residence), and whether the respondent reported a health problem; and model 3, added worry scales as additional mediators after adjusting for all the covariates in model 2. In accordance with our second hypothesis, we stratified our analyses according to the respondents’ age group: younger adults (20–39) and middle-aged adults (40–59). In addition to the main effects model described above, we also checked for interactions between age group and social role occupancy. All analyses in the present study were performed using the Statistical Analysis System, Release 8.02 (SAS Institute Inc., NC, USA).
Results Sample distribution (Table 1) After exclusion of households that did not provide complete information on relationships among family members, our final data set comprised 3011 females and 2682 males, aged 20–59. Women reporting ‘no social role’ consisted mainly of single young women in their 20 s, as well as students and other unpaid workers (excluding housewives). The corresponding category among men similarly consisted of single young men in their 20 s and students.
969
Number of social roles and health-related outcomes in women (Table 2) Among women overall, the risk of smoking declined as their number of social roles increased. We also found a significantly higher risk of smoking associated with moderate–high levels of worry about care-giving and human relations. The risk of drinking was nearly doubled among dual role women compared with single role women and the association was more pronounced in younger adults. We found a significantly higher risk of drinking habit associated with lower level of worry about care-giving and human relations. There was a highly significant inverse dose– response gradient between role occupancy and regular exercise, which was primarily observed among middleaged women (P-value for age interaction ¼ 0.07). Multiple social role occupancy was also associated with lower ORs of receiving a recent health check-up, as was having no social role. The OR of receiving a health check-up was also significantly reduced in women who expressed a moderate–high level of worry about care-giving and human relations. Contrary to the hypothesis that the assumption of multiple roles leads to greater role strain, we found no association between increasing role occupancy and self-reports of health problems among women in our sample. In fact, women reporting no roles (compared to single roles) were marginally significantly more likely to report an increased likelihood of reporting health problems (OR ¼ 2.21, 95% CI 0.90–5.44), although the direction of causality in this instance is not clear, i.e. sick women may choose not to take on multiple roles.
Number of social roles and health-related outcomes in men (Table 3) Stratifying the sample according to age group revealed an opposing trend among men: multiple role occupancy was associated with higher ORs of smoking among younger men, but lower ORs in middle-aged men (P-value for age role interactiono0.001). Paradoxically, we found a significantly higher risk of smoking associated with lower levels of worry about work, human relations and present finance, as well as a lower risk of smoking with moderate–high levels of worry about future health and finance. There was a strongly positive dose–response gradient between men’s role occupancy and drinking habit. Interestingly, that gradient appeared stronger among younger men (P-value for interaction between roles and age group ¼ 0.08).
ARTICLE IN PRESS 970
Y. Takeda et al.
Table 1 Distribution of demographic, health behaviours, socio-economic-related variables and social role occupancy Women
Number of respondents Age (years) 20–29 30–39 40–49 50–59 Smoking habit Smoker Ex-smoker Non-smoker Information missing Drinking habit Drinker Ex-drinker Non-drinker Information missing Exercise habit Yes No No (due to poor health condition) Information missing Had health check-ups in the previous year Yes No Information missing Have health problems Yes No Information missing Expenses per family member in the previous month o100 thousand yeny 100–200 200–300 4300 Information missing Domiciliary area of household o10 jyoz 10–30 30–50 50–70 470 Information missing Have paid job Yes No Marital status Married Unmarried Widowed Divorced Have dependent children (o20 years) in the household Yes No
Men
No.
(%)
No.
(%)
3011
(100.0)
2682
(100.0)
664 777 855 715
(22.1) (25.8) (28.4) (23.7)
574 704 794 610
(21.4) (26.2) (29.6) (22.7)
298 64 2014 635
(9.9) (2.1) (66.9) (21.1)
968 237 450 1027
(36.1) (8.8) (16.8) (38.3)
209 31 2135 636
(6.9) (1.0) (70.9) (21.1)
937 60 654 1031
(34.9) (2.2) (24.4) (38.4)
480 1772 131 628
(15.9) (58.9) (4.4) (20.9)
382 1222 48 1030
(14.2) (45.6) (1.8) (38.4)
1494 1421 96
(49.6) (47.2) (3.2)
1726 865 91
(64.4) (32.3) (3.4)
196 2734 81
(6.5) (90.8) (2.7)
165 2441 76
(6.2) (91.0) (2.8)
517 1741 465 145 143
(17.2) (57.8) (15.4) (4.8) (4.7)
457 1566 402 126 131
(17.0) (58.4) (15.0) (4.7) (4.9)
59 1255 1141 407 128 21
(2.0) (41.7) (37.9) (13.5) (4.3) (0.7)
96 1095 1009 349 111 22
(3.6) (40.8) (37.6) (13.0) (4.1) (0.8)
1617 1394
(53.7) (46.3)
2510 172
(93.6) (6.4)
2318 542 64 87
(77.0) (18.0) (2.1) (2.9)
1973 651 16 42
(73.6) (24.3) (0.6) (1.6)
1454 1557
(48.3) (51.7)
1307 1375
(48.7) (51.3)
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours
971
Table 1 (continued ) Women
Have dependent parents (470 years) in the household Yes No Number of social roles No roley Single rolez Double roles Triple rolesyy Quadruple roles(( Combination of social roles Employment role Employment and marital role Employment, marital and parenting role Employment, marital, parenting and caregiver role Marital role Marital and parenting role Marital, parenting and caregiver role No role Other combinations
Men
No.
(%)
No.
(%)
486 2525
(16.1) (83.9)
334 2348
(12.5) (87.5)
129 898 1172 615 197
(4.3) (29.8) (38.9) (20.4) (6.5)
118 554 649 1172 189
(4.4) (20.7) (24.2) (43.7) (7.0)
472 373 422 197 410 672 114 129 222
(15.7) (12.4) (14.0) (6.5) (13.6) (22.3) (3.8) (4.3) (7.4)
518 571 1092 189 27 14 3 118 150
(19.3) (21.3) (40.7) (7.1) (1.0) (0.5) (0.1) (4.4) (5.6)
‘Having a health problem’ was defined as ‘having a subjective health problem affecting daily life, having nursing care at home or
being hospitalized’. 94.0 yen was equivalent to US$1.00 at the 1995 average exchange rate. z 1 jyo is Japanese area unit: equivalent to 1.653 m2. y ‘No role’ means having no social role in the household, namely no paid job, not married, no dependent child(ren) nor dependent parent(s). z ‘Single role’ means playing just one social role in her/his household. ‘Double role’ means playing two social roles in her/his household. yy ‘Triple roles’ means playing three social roles in her/his household. (( ‘Quadruple roles’ means playing four social roles in her/his household. y
Men with higher role occupancy were more likely to report recent health check-ups than men with no or single role. Finally, men with no role had double the risk of reporting a health problem compared to the reference group, and middle-aged men were more likely to report a health problem compared with younger men. Increasing number of roles was associated with lower ORs of reporting health problems and this inverse gradient was particularly marked among middle-aged men (P-value for interactiono0.01). There was a significantly higher risk of reporting a health problem associated with worry about work, human relations and present finance, as well as worry about future health and finance.
Discussion Studies of multiple role occupancy in Western countries have failed to confirm earlier fears that women’s increasing participation in the paid labour
force would lead to role overload and increase their burden of mental and physical morbidity.4 Considerable evidence has supported the validity of the role expansion hypothesis, which suggests that the benefits of added roles for women have outweighed the costs.2,3 Fewer studies have been carried out among men, or directly compared men to women. When gender differences in the impact of multiple roles have been examined, the available evidence suggests that, in spite of the rise in dual earner couples, women continue to bear a disproportionate share of household responsibilities, resulting in the phenomenon of the ‘double shift’.17,18 Additionally, studies have demonstrated that the social context of roles matters in addition to the number of roles occupied by men and women.19,20 Few studies have been published on multiple role occupancy and health in Japan. The country, however, presents an interesting case study of a society undergoing transition. In recent decades, it has moved from a society with highly fixed and traditional gender roles, to an increasingly westernized
ARTICLE IN PRESS 972
Y. Takeda et al.
Table 2 Odds ratios and 95% confidence intervals for women’s health behaviours and health problems with the number of social roles
Age Younger adults (20–39 years) Middle-aged adults (40–59 years) Number of social roles in total age groups (20–59 years) No role Single role Double roles Triple roles Quadruple roles Worry level based on Future health and finance worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Care-giving and human relations worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Number of social roles in younger adults (20–39 years) No role Single role Double roles Triple roles Quadruple roles Number of social roles in middle-aged adults (40–59 years) No role Single role Double roles Triple roles Quadruple roles
Smoking
Drinking
Exercise
Health check-ups Health problems
OR
OR
OR
OR
95% CI
95% CI
95% CI
95% CI
OR
95% CI
1.00
1.00
1.00
1.00
1.00
0.58
0.44–0.76 1.27
0.92–1.75 1.57
1.25–1.97 1.80
1.53–2.11 1.81
1.26–2.62
0.90 1.00 0.98 0.67 0.27
0.45–1.81 1.52 1.00 0.72–1.33 1.70 0.46–0.99 1.56 0.12–0.61 1.24
0.61–3.80 0.88 1.00 1.14–2.53 0.77 0.98–2.49 0.61 0.58–2.64 0.58
0.46–1.67 0.34 1.00 0.60–0.99 0.50 0.44–0.83 0.66 0.36–0.94 1.11
0.22–0.51 2.06 1.00 0.41–0.61 0.82 0.53–0.83 1.07 0.78–1.57 0.50
1.01–4.23
1.00
1.00
1.00
1.00
1.00
0.95
0.63–1.44 0.74
0.45–1.24 0.93
0.65–1.33 1.17
0.90–1.53 7.80
5.40–11.3
1.45
0.85–2.48 0.69
0.33–1.42 1.10
0.68–1.76 1.20
0.84–1.71 5.47
3.33–8.98
1.00
1.00
1.00
1.00
1.00
1.24
0.88–1.77 1.67
1.15–2.43 0.91
0.68–1.23 0.91
0.74–1.13 0.99
0.64–1.52
1.57
1.07–2.29 1.20
0.75–1.91 0.73
0.50–1.07 0.72
0.56–0.94 1.31
0.81–2.12
0.96 1.00 1.16 0.80 0.36
0.44–2.08 1.72 1.00 0.74–1.80 2.48 0.47–1.37 2.13 0.13–0.95 1.19
0.57–5.18 1.12 1.00 1.25–4.93 1.02 0.96–4.74 0.94 0.30–4.63 0.90
0.54–2.33 0.32 1.00 0.66–1.59 0.35 0.56–1.57 0.55 0.42–1.94 1.46
0.20–0.50 2.21 1.00 0.26–0.46 1.02 0.40–0.76 1.30 0.88–2.41 0.35
0.90–5.44
1.24 1.00 0.82 0.55 0.17
0.14–11.0 1.76 1.00 0.53–1.27 1.38 0.31–0.97 1.34 0.04–0.73 1.26
0.21–15.1 0.70 1.00 0.84–2.26 0.65 0.75–2.39 0.47 0.50–3.14 0.44
0.14–3.46 0.11 1.00 0.48–0.89 0.67 0.32–0.71 0.82 0.24–0.82 0.90
0.02–0.53 2.07 1.00 0.51–0.87 0.73 0.60–1.12 0.98 0.56–1.47 0.53
0.55–1.22 0.67–1.70 0.20–1.24
0.49–2.13 0.56–2.99 0.04–2.90
0.47–9.05 0.46–1.17 0.56–1.72 0.19–1.48
Only parts of model 3 results for each outcome are shown.
pattern of active female labour force participation and shared household responsibilities. The present study sought to examine gender and age differences
in the impact of multiple role occupancy on healthrelated outcomes in the context of major societal changes, including the collapse of the lifetime
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours
973
Table 3 Odds ratios and 95% confidence intervals for men’s health behaviours and health problems with the number of social roles
Age Younger adults (20–39 years) Middle-aged adults (40–59 years) Number of social roles in total age groups (20–59 years) No role Single role Double roles Triple roles Quadruple roles Worry level based on Work, human relations and financial worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Future health and finance worry scale No worry (score ¼ 0) Low worry (score ¼ 1) Moderate–high worry (scoreX2) Number of social roles in younger adults (20–39 years) No role Single role Double roles Triple roles Quadruple roles Number of social roles in middle-aged adults (40–59 years) No role Single role Double roles Triple roles Quadruple roles
Smoking
Drinking
Exercise
Health check-ups Health problems
OR
OR
OR
OR
95% CI
95% CI
95% CI
OR
95% CI
1.00
1.00
1.00
1.00
1.00
1.06
0.81–1.39 1.64
1.29–2.08 0.91
0.69–1.20 1.01
0.83–1.23 2.52
1.60–3.96
0.35 1.00 0.98 1.26 1.28
0.19–0.64 0.28 1.00 0.68–1.41 1.68 0.91–1.76 2.08 0.73–2.23 2.08
0.13–0.59 0.87 1.00 1.20–2.34 1.11 1.54–2.80 0.78 1.28–3.36 0.58
0.45–1.69 0.53 1.00 0.77–1.60 1.62 0.56–1.09 1.88 0.31–0.99 1.19
0.35–0.81 2.22 1.00 1.24–2.11 0.57 1.49–2.37 0.30 0.81–1.74 0.37
1.07–4.62
1.00
1.00
1.00
1.00
1.00
1.57
1.13–2.19 1.12
0.84–1.49 0.98
0.72–1.34 1.09
0.87–1.36 1.47
0.93–2.33
0.86
0.61–1.21 0.91
0.67–1.24 0.78
0.54–1.13 1.02
0.79–1.30 1.85
1.15–2.97
1.00
1.00
1.00
1.00
1.00
1.20
0.74–1.93 0.93
0.61–1.41 0.66
0.39–1.11 1.21
0.86–1.72 7.40
4.73–11.6
0.52
0.27–1.00 0.81
0.45–1.47 1.42
0.74–2.71 0.88
0.56–1.40 6.44
3.68–11.3
0.36 1.00 1.28 1.68 4.52
0.19–0.68 0.29 1.00 0.77–2.13 1.39 1.14–2.50 2.33 1.01–20.5 2.91
0.13–0.64 0.99 1.00 0.87–2.21 1.04 1.63–3.33 0.90 1.05–8.10 0.60
0.50–1.98 0.54 1.00 0.61–1.77 2.03 0.60–1.37 1.77 0.17–2.20 0.75
0.35–0.84 2.74 1.00 1.37–3.01 1.20 1.34–2.34 0.61 0.33–1.67 1.41
1.11–6.79
y
0.26 1.00 0.18–0.84 1.72 0.21–1.00 1.75 0.16–1.00 1.80
0.03–2.56
0.51 1.00 0.51–1.82 1.22 0.33–1.18 1.61 0.20–1.06 1.05
0.10–2.58 3.76 1.00 0.76–1.97 0.31 1.01–2.59 0.15 0.59–1.86 0.20
1.00 0.39 0.46 0.40
Models did not converge due to small numbers.
y
1.00 0.93–3.19 0.97 0.95–3.22 0.62 0.86–3.79 0.46
Only parts of model 3 results for each outcome are shown. y
95% CI
0.33–0.98 0.18–0.52 0.16–0.86
0.47–3.05 0.28–1.32 0.17–11.7
0.65–21.6 0.16–0.60 0.07–0.30 0.08–0.52
ARTICLE IN PRESS 974 employment guarantee (and with it, Japanese men’s identity as primary bread-winners), as well as Japanese women’s rising dissatisfaction with traditional gender norms.
Multiple roles and health behaviours among younger (20–39) men and women The increased risks of both smoking and drinking among younger men with multiple roles seem to support our initial hypothesis about the effects of stress associated with juggling the multiple demands of work and family, particularly in the context of rising job insecurity and increasing expectations to share in household responsibilities. However, the finding that multiple role occupancy was associated with decreased risk of smoking among younger women was unexpected, particularly since the prevalence of smoking among young Japanese women has been rising rapidly in recent years.21 We offer two alternative explanations to account for this paradoxical finding. First, it is possible that young Japanese women are beginning to reap the rewards of role expansion, i.e. their increasing participation in the labour market has increased their status and bargaining position within the family and led to a corresponding decline in their stresscoping behaviour. Alternatively, the protective effect of multiple role occupancy on smoking among women may be an expression of traditional control over women’s behaviour. For example, women living with in-laws or children may be more likely to refrain from smoking due to family disapproval. At the same time, however, the higher prevalence of drinking among younger women with multiple roles seems to suggest a stress-coping response. In more detailed analyses (data not shown), the most significant factor associated with drinking among women was being in a parenting role. The gender discrepancy in regular health checkups, i.e. the strong inverse association with multiple role occupancy among younger women, but an association in the opposite direction for younger men, may be explained by differential access to work-based screening programs, rather than by an effect of multiple role occupancy per se (e.g. women not being able to have a check-up due to time pressures).22 In summary, multiple role occupancy among younger Japanese adults was simultaneously associated with both protective and damaging health behaviours. The net effect of these opposing
Y. Takeda et al. trends on future health outcomes among younger Japanese men and women remains to be determined.
Multiple roles and health behaviours among middle-aged (40–59) men and women Middle-aged women with multiple roles were less likely to smoke, but so were middle-aged men. Middle-aged women with multiple roles were no more likely to be drinkers compared to women with fewer roles, whilst we found scant evidence of an association between multiple role occupancy and classical stress-related coping behaviours among middle-aged men. This contrasts with the picture among younger men. It is unlikely that this difference among men reflects any difference in stress at work, since the recession in Japan (and the accompanying rise in job insecurity) affected both younger and older men alike. Instead, this difference may reflect changing expectations of gender roles, specifically for shared household responsibilities between younger women and men. In contrast to younger women, middle-aged women with multiple roles showed a strong inverse gradient in regular exercise, consistent with time constraints to invest in their own health. Overall, the protective effects of multiple roles on smoking appeared to cancel out the adverse effects on exercise, so that there was no association between multiple roles and self-reported health problems in middle-aged women. By contrast, occupying multiple roles appeared to have a net benefit for middleaged men, such that a strong inverse gradient was observed between the number of roles and the ORs of reporting health problems. In summary, we found strong evidence in support of the role expansion hypothesis for middle-aged Japanese men, i.e. the more roles they occupied, the healthier they seemed. Our study was limited in several respects. Firstly, it was cross-sectional in design and hence limited in its ability to offer causal interpretations. Thus, reverse causation cannot be excluded, in which healthy individuals were able to take on more social roles. Secondly, our study examined health-related behaviours, but not health outcomes as ultimate endpoints. Health problem(s) was a crude representation of health status, at best. Thirdly and most importantly, we lacked data on the quality of the different roles. Focusing on role occupancy per se may mask differences in role quality in each situation.20 Nor were we able to test more sophisticated theoretical models, such as
ARTICLE IN PRESS The impact of multiple role occupancy on health-related behaviours family-to-work spillover, or conversely, work-tofamily spillover.23 Notwithstanding these limitations, our study found intriguing gender and age differences in the relationships between multiple roles and health behaviours, within a large, representative sample of a society in transition. The economic and demographic trajectory of Japanese society suggests that the gendered division of labour in traditional households will not be tenable or sustainable in the future. Although our ‘snapshot’ of Japanese adults presents a mixed picture of the influence of multiple roles on the health of women and men, the ultimate balance of health-damaging and health-protecting influences will depend upon how gender relationships evolve within that society.
Acknowledgements This study was supported by Grant No. H11-toukei002 (Principal investigator of the national research group: Prof. Hiroshi Yanagawa, President of Saitama Prefectural University) from the Ministry of Health, Labour and Welfare, Japan. Statement of competing interest: There are no competing interests on this paper.
References 1. Goode WJ. A theory of role strain. Am Sociol Rev 1960;25:483–96. 2. Marks SR. Multiple roles and role strain: some notes on human energy, time and commitment. Am Sociol Rev 1977;42:921–36. 3. Sieber SD. Toward a theory of role accumulation. Am Sociol Rev 1974;39:567–78. 4. Walsh DC, Sorensen G, Leonard L. Gender, health, and cigarette smoking. In: Amick III BC, Levine S, Tarlov AR, Walsh DC, editors. Society and health. New York: Oxford University Press; 1995. p. 131–71. 5. Rodin J, Ickovics JR. Women’s health: review and research agenda as we approach the 21st century. Am Psychol 1990;45:1018–34.
975
6. Thoits PA. Multiple identities and psychological well-being: a reformulation and test of the social isolation hypothesis. Am Sociol Rev 1983;48:174–87. 7. Barnett RC, Baruch GK. Social roles, gender, and psychological distress. In: Barnett RC, Biener L, Baruch GK, editors. Gender and stress. New York: The Free Press; 1987. p. 122–43. 8. Cleary PD, Mechanic D. Sex differences in psychological distress among married people. J Health Soc Behav 1983;24:111–21. 9. United Nations Development Program. Human Development Report 1999: globalization with a human face. New York: Oxford University Press; 1999. 10. National Institute of Population and Social Security Research, Japan. The Second National Survey on Family in Japan; 1998. 11. Cabinet Office, Japan, Gender Equality Bureau. The Public Opinion Survey on a Gender-Equal Society of Japan; 1992, 1997, 2002. 12. Horlacher DE. Aging in Japan: causes and consequences. International Institute for Applied Systems Analysis. Laxenburg, Austria: IIASA; 2002. 13. Ministry of Health, Labour and Welfare, Japan, Statistics and Information Department. The 1995 Comprehensive Survey of the Living Conditions of People on Health and Welfare (CSLCPHW); 1995. 14. Ministry of Health, Labour and Welfare, Japan, Health Service Bureau. The 1995 National Nutrition Survey (NNS); 1995. 15. Hashimoto S, Kawado M, Matsumura Y, et al. Hoken-toukei ni okeru record-linkage no jisshi kanousei (The feasibility study of record-linkage on Japanese health statistics). J Health Welfare Stat 2001;48:1–5 [in Japanese]. 16. Kawachi I, Kennedy BP. The relationship of income inequality to mortality: does the choice of indicator matter? Soc Sci Med 1997;45:1121–7. 17. Hochschild AR. The second shift. New York: Viking; 1989. 18. Bird CE, Fremont AM. Gender, time use and health. J Health Soc Behav 1991;32:114–29. 19. Moen P, Dempster-McClain D, Williams Jr RM. Social integration and longevity: an event history analysis of women’s roles and resilience. Am Soc Rev 1989;54:635–47. 20. Griffin JM, Fuhrer R, Stansfeld SA, et al. The importance of low control at work and home on depression and anxiety: do these effects vary by gender and social class? Soc Sci Med 2002;54:783–98. 21. Honjo K, Kawachi I. Effects of market liberalisation on smoking in Japan. Tob Control 2000;9:193–200. 22. Ministry of Public Management, Home Affairs, Posts and Telecommunications, Japan, Statistics Bureau. The 2002 Labor Force Survey; 2002. 23. Barnett RC, Baruch GK. Women’s involvement in multiple roles and psychological distress. J Pers Soc Psychol 1985;49: 135–45.