Work and family roles and the association with depressive and anxiety disorders: Differences between men and women☆

2.1. Research population 

The Netherlands Mental Health Survey and Incidence Study (NEMESIS) is an epidemiological study in the Dutch general population to determine the prevalence of psychiatric disorders. A representative sample of adults, aged 18–64 years, was interviewed by intensively trained and monitored interviewers in 1996 (n=7076), (Bijl et al., 1998). Psychiatric disorders were measured using a computerized version of the Composite International Diagnostic Interview (CIDI, version 1.1). The CIDI is a validated diagnostic instrument, developed by the World Health Organization (WHO), which is suitable for non-clinician interviewers (Robins et al., 1988). For this study, we explored the association of social roles (employee, partner, and parent) with the 12-month prevalence of depression and anxiety disorders. We selected people with theoretical ability to have the three roles of being a partner, a parent and having a job. Because people younger than 25 may still be students and people over 55 may be retired, we selected people in the age between 25 through 55 years (n=5227), the age group in which the occurrence of the three social roles is most likely. We excluded all respondents (n=1219) with any psychiatric diagnosis (anxiety, depression, dysthymia, psychosis, bipolar disorder, eating disorder, or substance dependence) before the last 12 months, since we hypothesized that having a lifetime diagnosis of a psychiatric disorder could have influenced the development of social roles. For the same reason, we excluded respondents who received a social security benefit because they were declared unfit to work (n=151). The remaining 3857 persons, 1925 women and 1932 men, constitute the sample for the present study.

2.2. Depression and anxiety disorders 

The dependent variable of this study was the 12-month prevalence of depression and anxiety disorders, defined and measured using DSM-III-R criteria. Depressive disorders included major depression and dysthymia; anxiety disorders included social phobia, panic disorders, simple phobia and generalized anxiety disorder. We considered all these disorders as one group, since these disorders have several shared characteristics in their origins and treatment and co-morbidity among these disorders is high (Boyer, 2000, Himmelhoch et al., 2001, Levine et al., 2001).

2.3. Work-role variables 

The variable work (1=paid work, 0=no paid work) was constructed based on a question about one's current sources of income with 12 categories. The four categories ‘paid work’, ‘part-time job’, ‘temporary work or short-term sick leave’ or ‘household and part-time work’ were recoded ‘working’. All other categories, such as ‘looking for work’, ‘studying’, ‘voluntary work’ and ‘household’ were recoded ‘not working’. In line with previous analyses (Plaisier et al., 2007), we recoded persons who reported to work less than 8 h a week as not working. Based on the question ‘In the last four weeks, how many hours a week have you been working on average?’ a variable ‘amount of working hours’ was created with three categories: 1=0 through 7 h (not working); 2=8 through 35 h (part-time job); 3=36 h or more a week (full-time job). These categories were based on Dutch standards, which consider 36 h a week as full-time work.

2.4. Family-role variables 

The partner role was defined by a dichotomized variable partner (1=living with partner, 0=not living with partner). The parent role was defined by 4 variables: child (1=has child; 0=has no child), child in household (1=has child in household; 0=has no child in household), the age of the respondent at which his or her first child was born, the number of children and the age group of the youngest child. The age categories of children were based on main stages in development: 1=0 to 3 years (baby/toddler, not of school age); 2=4 to 11 years (primary school age); 3=12 to 17 years (adolescent, secondary school); 4=18 to 24 years (young adult, studying or working in first jobs); 5=25 years or older (independent adult).

2.5. Social roles 

Based on the dichotomous work and family variables a variable for number of social roles was constructed with a range from 0 (no roles) to 3 (all three roles). Beside this, we computed interaction terms of the combinations of social roles: (1) work and child, (2) work and partner, (3) partner and child and (4) work and partner and child.

2.6. Confounding variables 

Confounding variables were age, level of education (1=low (no qualifications); 2=lower intermediate (low vocational or secondary school); 3=higher intermediate (intermediate vocational or high school); 4=high (high vocational or academic)) and the number of chronic diseases. For this last variable, the number of chronic diseases for which the respondent had medical treatment was counted (up to 31 diseases, e.g. lung diseases, cardiovascular diseases, cancer, diabetes, rheumatoid arthritis, kidney or liver diseases and muscular diseases).

2.7. Statistical analyses 

With independent t-tests (for continuous variables) and chi-square tests (for categorical variables), the gender differences in work and family roles, and socio-demographic variables were examined (Table 1). To address the first hypothesis, we tested the association of social roles and gender differences with the presence of a depressive or anxiety disorder in a logistic regression model, controlling for age, level of education and the number of chronic diseases (Table 2). Odds ratios and the 95% confidence intervals were calculated for men and women separately. The associations of social role combinations with present anxiety and depressive disorders (second hypothesis) were tested in a controlled model with the individual social roles, by adding the interaction term between these social roles (Table 3). We tested whether associations between social roles and depression and anxiety were different for men and women by testing gender and social role interaction terms in adjusted logistic models. These interactions were tested at a significance level of 95% (p < 0.05). To explore whether social roles contribute to the female preponderance in depression and anxiety disorders, the reduction of the OR of gender for the prevalence of depression and anxiety disorders was calculated after adding work and family variables into the adjusted logistic model (Table 4). Reductions of more than 10% were considered as indications of an essential change in risk. For this study, SPSS 11.0 was used to perform the analyses.

Table 1. The prevalence of depression and anxiety disorders, socio-demographic variables and work and family roles among men and women aged 25 through 55 years

		Total	Men	Women	Gender difference
		N=3857	N=1932	N=1925	(p-value)
	12-month prevalence of depression and anxiety disorders (%)	12.5	8.8	16.3	<0.001
	Socio demographic variables
	Age	38.8 (±8.31)	38.9 (±8.21)	38.7 (±8.42)	0.49
	Level of education (%)				<0.001
	Low	3.8	3.4	4.2
	Lower intermediate	36.8	31.0	42.5
	Higher intermediate	29.1	30.3	27.8
	High	30.4	35.2	25.5
	Number of physical illnesses	0.70 (±0.99)	0.54 (±0.84)	0.87 (±1.11)	<0.001
	Family roles
	Partner (% yes)	77.1	76.4	77.9	0.28
	Children (% yes)	68.3	63.5	73.2	<0.001
	Children in household (% yes)	56.4	51.7	61.1	<0.001
	Number of children (%)				<0.001
	0	31.7	36.5	26.8
	1–2	48.6	43.8	53.4
	≥3	19.8	19.7	19.8
	Age at birth first child (n=2534)	26.9 (±4.5)	28.2 (±4.5)	25.9 (±4.2)	<0.001
	Age group youngest child (%, n=2634)				0.01
	0–3	25.9	26.6	25.3
	4–11	32.0	34.2	30.1
	12–17	17.2	17.1	17.3
	18–24	17.4	16.7	18.0
	≥25	7.4	5.3	9.3
	Work role
	Work (% yes)	80.8	95.0	66.5	<0.001
	Hours of work per week (%)				<0.001
	Not working (<8 h)	19.2	5.0	33.5
	8–35	26.2	8.3	44.2
	≥36	54.6	86.7	22.4
	Role combinations
	Combinations of work, child, partner (%)				<0.001
	Not working, no child, no partner	1.8	2.3	1.2
	Not working, no child, has partner	1.1	0.7	1.6
	Not working, has child, no partner	2.9	0.5	5.4
	Not working, has child, has partner	15.6	1.7	29.5
	Working, no child, no partner	13.7	16.4	11.0
	Working, no child, has partner	15.1	17.1	13.0
	Working, has child, no partner	4.5	4.5	4.6
	Working, has child, has partner	45.3	56.9	33.7
	Number of social roles (work, child, partner %):				<0.001
	0 roles	1.8	2.3	1.2
	1 role	17.7	17.5	18.0
	2 roles	35.2	23.3	47.1
	3 roles	45.3	56.9	33.7

Table 2. The adjusted association between work and family roles and the prevalence of depression and anxiety disorders among men and women aged 25 through 55 yearsa

		Men	p-value	Women	p-value	gender interaction
		OR (95% CI)		OR (95% CI)		(p-value)
	Family roles
	Partner (0=no, 1=yes)	0.63 (0.44–0.90)	0.01	0.45 (0.35–0.60)	<0.001	0.13
	Children (0=no, 1=yes)	0.75 (0.51–1.09)	0.13	0.99 (0.73–1.36)	0.96	0.56
	Children in household (0=no, 1=yes)	0.87 (0.62–1.20)	0.38	1.07 (0.83–1.38)	0.61	0.37
	Number of children		0.049		0.25	0.06
	0	Reference group		Reference group
	1–2	0.84 (0.57–1.24)	0.38	0.93 (0.68–1.29)	0.68
	≥3	0.52 (0.30–0.88)	0.02	1.22 (0.82–1.81)	0.31
	Age at birth first child	1.03 (0.98–1.08)	0.22	1.01 (0.98–1.05)	0.48	0.40
	Age group youngest child		0.37		0.32	0.62
	0–3	0.62 (0.17–2.21)	0.46	1.59 (0.63–3.99)	0.33
	4–11	1.15 (0.39–3.34)	0.81	1.97 (0.93–4.21)	0.08
	12–17	1.08 (0.39–2.98)	0.89	1.67 (0.87–3.23)	0.13
	18–24	0.89 (0.33–2.40)	0.82	1.56 (0.85–2.87)	0.15
	Work role
	Work	0.40 (0.24–0.69)	0.001	0.86 (0.66–1.12)	0.23	0.02
	Hours of work a week		<0.001		0.36	0.05
	<8 h	Reference group		Reference group
	8–35 h	0.86 (0.43–1.73)	0.69	0.94 (0.71–1.25)	0.67
	≥36 h	0.39 (0.22–0.69)	0.001	0.77 (0.53–1.11)	0.16

a Adjusted for age, level of education and number of chronic diseases.

Table 3. The adjusted association between role combinations and the prevalence of depression and anxiety disorders among men and women aged 25 through 55 yearsa

	Role combinations:	Men		Women
		OR (95% CI)	p-value	OR (95% CI)	p-value
	Number of social roles:		0.01		<0.001
	0 roles	Reference group		Reference group
	1 role	0.50 (0.22–1.14)	0.10	0.44 (0.18–1.11)	0.08
	2 roles	0.36 (0.16–0.81)	0.01	0.29 (0.12–0.70)	0.01
	3 roles	0.30 (0.14–0.67)	0.001	0.23 (0.09–0.58)	0.001
	Work ⁎ Partner	1.18 (0.39–3.61)	0.78	1.11 (0.63–1.95)	0.72
	Work ⁎ Child	1.49 (0.49–4.56)	0.48	3.15 (1.56–6.37)	0.001b
	Partner ⁎ Child	0.84 (0.37–1.92)	0.86	0.50 (0.27–0.90)	0.02 c
	Work ⁎ Partner ⁎ Child	4.19 (0.34–52.2)	0.27	0.64 (0.15–2.75)	0.54

a Adjusted for age, level of education, number of chronic diseases, and also individual roles in analyses examining interaction effects. b Among women without children the OR of work=0.28 (95% CI: 0.14–0.54, p<0.001), among women with children the OR of work=1.01 (95% CI: 0.75–1.35, p=0.95). c Among women without children the OR of having no partner=1.71 (95% CI: 1.02–2.86, p=0.04) among women with children the OR of having no partner=3.04 (95% CI: 2.15–4.32, p=0.001).

Table 4. The contribution of work and family roles to the gender difference in the prevalence of depression and anxiety disorders among men and women aged 25 through 55 yearsa

		Odds ratio (95% CI)	p-value	Reduction in gender OR
	Gender difference	1.71 (1.40–2.10)	<0.001
	Gender difference adjusted for
	Partner (0=no, 1=yes)	1.74 (1.42–2.14)	<0.001	0%
	Child (0=no, 1=yes)	1.73 (1.41–2.13)	<0.001	0%
	Work (0=no, 1=yes)	1.56 (1.25–1.94)	<0.001	21.1%
	Hours of work a week	1.25 (0.96–1.63)	0.10	64.8%
	Number of social roles	1.69 (1.37–2.09)	<0.001	2.8%
	Partner ⁎ Child	1.62 (1.32–1.99)	<0.001	12.7%
	Partner ⁎ Work	1.61 (1.29–2.01)	<0.001	14.1%
	Work ⁎ Child	1.67 (1.34–2.09)	<0.001	5.6%
	Work ⁎ Partner ⁎ Child	1.63 (1.30–2.04)	<0.001	11.3%

a Adjusted for age, level of education and number of chronic diseases. When examining interaction effects, individual roles were also included.

3.1. Sample description 

Table 1 shows a significantly higher 12-month prevalence of depression and anxiety disorders in women (16.3%) than in men (8.8%, χ2=48.99, p<0.001). In this sample, 483 persons had an anxiety or depressive disorder from whom 242 (50%) had a prevalent anxiety disorder, 136 persons (28%) had a prevalent depressive disorder and 105 persons (22%) had both. Men were higher educated and had fewer chronic diseases. The proportion of men and women living with a partner did not differ. Women were more likely to be a parent, had more children, and were younger when they became parent. The most pronounced differences between men and women occurred in work roles. More men than women had a paid job, and men worked more hours a week compared to women. There were also differences between men and women in their combinations of social roles. Though the majority of both men and women combined work with parenthood and partnership, among women the combination of partner and parenthood without a paid job was also common. The number of social roles differed between men and women; more men than women had three social roles (56.9% against 33.7%) and women occupied more frequently two roles (47.1% against 23.3% among men).

3.2. Social roles and the association with depression and anxiety disorders 

Table 2 presents the odds ratios for depression and anxiety disorders associated with social roles, after adjustment for age, level of education and diseases. Living with a partner was associated to a decreased odds ratio of depression and anxiety disorders consistently for both men and women. Having children was not associated with the depression and anxiety disorders, nor has the respondent's age at birth of the first child or the age group of the youngest child. None of the gender interaction terms of these parent-role variables, added into the adjusted model, was significant. The number of children was only significant among men. Having three children or more seemed to decrease the a prevalent depressive and anxiety disorder (OR=0.52, 95% CI: 0.98–1.08, p=0.02). In contrast, among women the OR of depressive and anxiety disorder associated with having three children or more was 1.22 (0.82–1.81, p=0.48). The p-value of the gender interaction of number of children added into the adjusted model was low, but not significant (p=0.06). A more pronounced gender difference was found for the work-role variables. Having a job was a strong protective factor of the presence of depressive and anxiety disorders for men, but not for women. This gender difference was supported by the fact that the interaction term for gender was significant (p=0.02). A gender difference was also significant for the hours of work per week (p=0.046). Particularly full-time work was associated with a better mental health among men (OR=0.39, 95% CI: 0.22–0.69, p=0.001). Although the associations of work and hours of work a week with anxiety and depressive disorders among women were in the same direction, these associations were not significant.

3.3. Role combinations and the association with depression and anxiety disorders 

Table 3 presents the associations of social role combinations with the prevalence of depression and anxiety among men and women separately. After adjustment for age, level of education and number of chronic diseases, having more social roles was associated with a decreased prevalence of depression and anxiety disorders among both men and women. Adding the three individual roles of work and family (being a partner and being a parent) into this adjusted model outweighed the effect of the multiple social roles variable, indicating that the effect of having more social roles does not have a supplementary effect beside direct effects of the individual social roles (specific results not shown). Nevertheless, we found some gender differences in the effect of particular combinations of social roles. The interaction term of work and child added into the adjusted model with the individual variables was significant among women (p=0.001), but not among men (p=0.45). This indicates that the combination of the work role with parenthood was associated to the presence of anxiety or depression disorders among women, but not among men. The OR of depression and anxiety disorders for having a job was 0.28 (95% CI: 0.14–0.54, p<0.001) among women without children and 1.01 (95% CI: 0.75–1.35, p=0.95) among women with children. The protective effect of the work role seems only significant among women without children. For men, the protective effect of the work role was significant whether they had a child or not. We also found a significant interaction effect for the combination of the partner role with the parent role among women (p=0.02), which was not significant among men (p=0.86). For women, the partner role was more strongly associated with the presence of depression and anxiety disorders among those with children compared to those without children. The OR of anxiety and depression associated with having no partner was 3.04 (95% CI: 2.15–4.32, p=0.001) among women with children and 1.71 (95% CI: 1.02–2.86, p=0.04) among women without children. No significant interaction effects for the combination of the work and partner roles were found, nor for the combination of the three roles (all p-values >0.05).

Because of the observed gender difference in the combination of the work role with the parent role, we also examined the interaction of hours or work with parenthood among men and women separately (results not shown). Among men this interaction term was not significant (p=0.68) but it was significant among women (p=0.005). For women without children, the ORs were 0.71 (95% CI: 0.30–01.81, p=0.47) for part-time jobs and 0.34 (95% CI: 0.16–0.71, p=0.004) for full-time work compared to women working less than 8 h. Among women with children the ORs were respectively 1.23 (95% CI: 0.42–3.61, p=0.72) for part-time jobs, and 0.54 (95% CI: 0.21–1.36, p=0.19) for full-time work. We did not observe differences between women working less than 16 h or 17 to 35 h a week (results not presented). This indicates that working more hours is advantageous for both men's and women's mental health, but having children mitigates this effect among women.

3.4. The explanation of gender differences in the prevalence of depressive and anxiety disorders 

Women were more likely to have depressive and anxiety disorders than men: the OR for women compared to men was 1.71 (95% CI: 1.40–2.10, p<0.001) adjusted for age, level of education and the number of chronic diseases. Table 4 shows changes in the OR for the prevalence of depression and anxiety disorders by gender, after adding various social role variables into the model. Adding work status changed the OR of gender from 1.71 to 1.56, a reduction of (1.71–1.56)/(1.71–1) ⁎ 100=21.1%. The largest reduction (64.8%) of the OR occurred when hours of work was added into the model. The OR of gender decreased to 1.25 (95% CI: 0.96–1.63, p=0.11) and was no longer significant. The partner and parent-role variables did not reduce the odds ratio of gender. The interaction term of work and partner also reduced the OR of gender considerably with 14.1%, but this was not more than the reduction caused by the individual work-role variable (21.1%). The three-way interaction-term work, child and partner reduced the OR of gender not more (11.3%) than the individual work-role variable, indicating that the main reducing effect was that of the work-role variable. On the other hand, family roles seem to mitigate the reducing effect of the work role on the OR of gender, illustrated by lower percentages of gender OR reductions by respectively the partner and work-role interaction (14.1%) and the parent- and work-role interaction (5.6%) compared to the contribution in the gender OR reduction by the work role individually (21.1%).