Total workload, work stress and perceived symptoms in Swedish male and female white-collar employees

Abstract

Background: The aim of this study was to analyse how paid work, unpaid household tasks, child care, work–child care interactions and perceived work stress are associated with reported symptoms in male and female white-collar employees. Methods: A questionnaire was mailed to 1300 men and 1300 women belonging to the white-collar sector, with at least 35 hours of regular employment a week and a participant age of between 32 and 58 years. It contained items relating to total workload (hours spent on paid work, unpaid household tasks and childcare), subjective indices for work stress and symptoms. The response rate was 65% (743 women; 595 men). Gender difference in symptom prevalence was tested by analyses of variance. Odds ratios were used to estimate the bivariate associations between work-related variables and symptom prevalence. A multivariate analysis estimated the effect of paid and unpaid work interaction, work–childcare interplay and possible synergy. Results: The frequency and severity of symptoms was higher in women than in men (P < 0.0001). Employed women's health was determined by the interaction between conditions at work and household duties (OR 2.09; 1.06–4.14), whereas men responded more selectively to long working hours, i.e. >50 h/week (OR 1.61; 1.02–2.54). However, childcare (<21 h/week) appeared to have a buffer effect on the risk of a high level of symptoms in men working long hours. Conclusion: Working life and private circumstances and the interplay between them need to be taken into account to curb stress-related ill health in both men and women.

Work-related health problems have increased for both men and women in industrialised countries.^1,2 Neck–shoulder pain, tiredness, headache, stomach problems, low back pain and sleeping problems are the most frequently-reported symptoms resulting in sickness absence.^1,3,4 Work-related illnesses in Sweden increased by two-thirds for women and almost doubled for men between 1996 and 1998.¹

Several models have been presented to explain the association between workload (paid and unpaid work), perceived stress and health complaints.^5–7 Higher levels of job stress have been reported in well-educated younger male and female employees working excessive overtime, and in high-level managers and administrators.^8,9 The pattern of major job stressors appears to vary by workload, employment grade, occupational level, age and gender.

To identify important health status determinants in women, the multiple roles of women have been investigated.^10–12 The multiple role concept—being a spouse, a mother and an employee—suggests that harmful effects in relation to health outcomes may be produced by role conflict and role overload.¹⁰ On the other hand, there is evidence that the benefits of assuming multiple roles could outweigh the possible harmful effects by providing improved social networks, financial independence and greater self-esteem, with one role compensating for the negative aspects of others.¹¹ There is supporting evidence for both models, but most studies indicate that women benefit from shouldering multiple roles.^13–15 One study found that shouldering multiple roles was beneficial for women without children and for women over 40 years of age with children, while for younger women working full-time and having small children was linked to role conflict, overload and poorer health.¹⁵ Fokkema also found that part-time work and having older children are associated with better health.¹⁶ Recent research confirms that both employment and marriage have beneficial effects on women's health.¹² However, much less is known about work–marriage–childcare combinations and health in men.

The high female participation in the work force and the persisting traditional gender role pattern with regard to home and family responsibilities thus seem to contribute to observed gender differences in work stress.^8,17,18

However, a deeper understanding of the association between workload (paid and unpaid), perceived stress and reported symptoms in both men and women, undertaken in the same study, is important in order to avoid unnecessary medication and sick-leave among women in the first case. The aim of this study was to analyse how paid work, unpaid household tasks, childcare and work–childcare interactions, as well as perceived work stress, are associated with reported symptoms in male and female white-collar employees in comparable life situations.

Methods

Selection of subjects

The selection of study participants was based on a previous study⁸ and was made according to the following criteria: belonging to the white-collar sector; at least 35 hours of regular employment a week and participant age between 32 and 58 years.

In 2001, samples of men and women were selected randomly, by Statistics Sweden, from four occupational areas in the white-collar sector: (1) technology and natural science (architects, engineers and technicians in chemistry, physics and construction), (2) education (university professors and lecturers, school principals and other faculty members, school teachers), (3) health care (physicians, nurses, dentists, psychologists) and (4) administrative work (company and personnel administrators, computer programmers and system operators).

From these areas, women and men were matched for age, occupational level and children below the age of 18 living at home. Male and female participants were not selected from the same families, although this may have happened in single cases by chance.

This resulted in a study cohort of 1025 men and 1025 women, with at least one child under 18 years of age living at home, and 275 men and 275 women without children at home. The following occupational levels were represented: (1) top managerial level, director, managing director, deputy executive; (2) upper middle managerial level, head of a large division with 11–50 employees; (3) lower middle managerial level, head of a small division with up to 10 employees; (4) semi-managerial level, e.g. foreman without formal personnel responsibilities; (5) no managerial duties.

A questionnaire containing items relating to socio-economic factors, total workload (TWL) and subjective indices for work stress and symptoms was mailed to the study participants and completed anonymously. For this reason, a reminder was sent by mail to everyone after three weeks. Incomplete questionnaires were excluded. Fifty respondents did not fulfil the criteria set for participation, as they were either on maternity leave, sick listed on a long-term basis, or studying. The statistical analyses were based on data from 743 women and 595 men (table 1), equivalent to a response rate of 65%.

The TWL instrument

Total workload encompasses paid and unpaid forms of productive activity, as described by Kahn.¹⁹ A person's TWL thus encompasses paid employment and overtime at work and unpaid work duties, such as household chores, childcare, care of elderly or sick relatives, work in voluntary organisations, unions and so on.

A Swedish TWL instrument, which was developed and psychometrically evaluated by Mårdberg and colleagues and was designed specifically for measuring aspects of TWL in female and male white-collar employees, was used in this study.²⁰ It contains scales for measuring paid job descriptors, personal control, workload and qualification in the different activities.

Calculation of TWL and work stress

Each individual's TWL was calculated by adding up self-reported data relating to the average number of hours a week devoted to (i) paid employment and overtime at work, (ii) household duties (mending, sewing, laundry, gardening), (iii) childcare (homework/teaching, care-taking, playing) and (iv) other unpaid duties (voluntary work in unions and organisations, care of sick or elderly relatives).

The items making up the four indices related to work stress, i.e. the subjective TWL indices, were calculated using mean raw scores (scale 1–7) and were as follows:²⁰ The indices ‘Stress from paid work’ and ‘Conflict between demands’ are subscales of ‘Perceived TWL’.²⁰ Consequently, these indices were correlated with ‘Perceived TWL’, 0.75 and 0.76 respectively. The correlation between ‘Perceived TWL’ and ‘Control over household work’ was low, −0.20.

1. Perceived TWL: too much to do on and off the job, overall stress, demands

   1. Stress from paid work: too much to do at work, stress, demands at work

   2. Conflict between demands: childcare and household chores contribute to TWL, conflict between duties

2. Control over household work: influence at home, control, opportunity to make own decisions

All the TWL variables and the subjective TWL indices were dichotomised at the higher quartile of the distribution.

Calculation of symptom prevalence

As an indicator of perceived health, the dependent variable was assessed by a number of symptoms that are commonly reported by women and men in population-based studies.^21–23 This procedure is in line with the findings from a study where the most common physical and psychological symptoms could be interpreted as a single entity phenomenon mirroring general distress.²⁴ The symptoms were stomach pain, headache, sleep disturbances, dizziness, low back pain, loss of appetite and shoulder and neck pain (Cronbach alfa 0.60). The frequency (never, every second week or every week) and severity (mild, moderate, severe) of the symptoms was reported. To calculate symptom prevalence, taking both frequency and severity into consideration, a variable was created in which the seven symptoms were taken together and dichotomised into those with a high level of symptoms, equal to suffering from at least two symptoms present every week or every second week of a moderate or severe character, as opposed to those with a low level of symptoms, characterised by no symptoms or only one symptom of any frequency and severity.

Statistical analyses

Men and women were compared throughout the study. Gender difference in symptom prevalence was tested by analyses of variance (ANOVA), fixed model. Odds ratios were used to estimate the bivariate associations between TWL variables and subjective TWL indices in relation to a high level of symptoms. Men and women were treated in the same analysis through introducing an interaction term between gender and each work-related variable.

The bivariate analyses were further supplemented by multiple logistic regression analyses to adjust for age, cohabitation status, number of children living at home and occupational level, as the matching of participants was not completely successful, but as hardly any difference in the risk of a high level of symptoms was detected, they are not presented.

To estimate the effect of combined exposure (i.e. paid and unpaid work interaction and work–childcare interplay), a multivariate analysis was performed with a combined variable. To further test for effect modification, a synergy index (SI) was calculated as proposed by Rothman.²⁵

The following algorithm was used to determine effect modification, where SI > 1 signifies a synergistic effect and SI < 1 an antagonistic effect.

SI=[RR(AB)−1]/{[RR(Ab)−1]+[RR(aB)−1]}where RR = risk ratio, Ab = exposed to one of the factors, aB = exposed to the other factor and AB = exposed to both factors. Odds ratios were used as estimates of relative risks.

The Statistical Package for the Social Sciences (SPSS) was used for all statistical purposes.²⁶ Statistical significance was determined at the 95% confidence interval (CI) level.

This project was approved by the Ethics Committee at Stockholm University.

Results

Socio-demographic data reveal that somewhat more men than women were in top level positions, while more women than men (19% women and 11% men) lived without a partner. Of these, 75 (53%) women and 11 (17%) men had at least one child younger than 18 years of age living at home (table 1).

Women had a higher TWL than men and spent more time on household work and childcare but fewer hours in paid work than men did. Furthermore, women obtained higher scores on all the subjective TWL indices than men and a statistically significant difference was obtained through analysis of variance, with P<0.0001.²⁷

Symptom frequency was higher in women than in men for all symptoms and, moreover, more women than men rated their symptoms as moderate or severe (table 2).

A mean was calculated for each symptom by analysing the constructed variable that combines the frequency and severity of each symptom (table 3). It was found that suffering from a particular symptom every week or every second week, of moderate or severe character, was more common in the women than among the men, and gender differences were statistically significant for five of the seven symptoms (P < 0.05). The most prevalent and severe symptom in the women was shoulder and neck pain, followed by headache and sleep disturbances, while sleep disturbances, low back pain and shoulder and neck pain were the most prevalent and severe symptoms among the men.

Analysing the combined variable, where all the symptoms were taken together, revealed that 30% of the women and 15% of the men suffered from two or more symptoms appearing every week or every second week, of moderate or severe character, with P < 0.0001 (table 3).

Of the TWL variables, i.e. those variables reflecting the actual time spent on the various activities, none was associated with an increased risk of a high level of symptoms among the women, while only ‘being in paid work more than 50 hours a week’ was associated with this risk among the men (table 4). However, 17% of the women and 32% of the men worked more than 50 hours a week, equivalent to overtime of more than 10 hours a week.

When analysing the subjective TWL indices, the strongest association was found for ‘stress from paid work’ for both women and men, with OR 3.54 (2.53–4.96) and OR 4.29 (2.67–6.91), respectively. Furthermore, ‘perceived total workload’ displayed a statistically significant risk of suffering from a high level of symptoms for both women (OR 1.93; 1.18–3.13) and men (OR 3.13; 1.61–6.08), while ‘conflict between demands’ and ‘control over household work’ did not. ‘Conflict between demands’ was further analysed specifically for women with children younger than seven and 11 years, respectively, but no statistically significant association with a high level of symptoms was found.

Finally, the association between paid work and a high level of symptoms was investigated when there was a concurrent exposure to a high level of household responsibility, i.e. a double exposure situation (table 5). This resulted in an odds ratio of 2.09 (1.06–4.14) and a synergy index of 3.7 for women, adjusted for age, indicating a considerably higher risk of symptoms as compared to being exposed to only one of the factors. For men, the double exposure situation was not associated with an increased risk of a high level of symptoms (OR 1.69; 0.46–6.21), but only 16 men were in this situation.

The work–child care interplay (table 5) was not associated with an increased risk of a high level of symptoms (OR 0.78; 0.16–3.88) among women, while an increased risk was noted for men (OR 2.96; 1.02–8.57), with synergy also being indicated (SI = 5.3), but few men were in this situation. However, in the case of those men spending more than 50 h/week in paid work, who also devoted a restricted amount of time to childcare (<21 h/week), a reduced risk of a high level of symptoms was noted, as the effect of paid work thus lost its statistical significance (OR 1.46; 0.73–2.93).

Discussion

The main aim of this study was to investigate how conditions related to paid and unpaid work and the interplay between them were associated with ill-health, in terms of a high level of common symptoms, in highly-educated white-collar men and women, aged between 32 and 58.

A considerable number of men and women were engaged in overtime and almost twice as many women as men were suffering from frequent and severe symptoms.

Employed women's reported symptoms were determined by an interaction between conditions at work and at home, whereas men responded more selectively to work conditions. On the other hand, a limited amount of childcare was able to reduce the observed symptom reporting from excessive paid work in this group of highly educated men. In line with this, Ozer and colleagues found that, the more fathers participated in childcare relative to their wife, the lower the father's distress.²⁸

Furthermore, the perception of stress from paid work was associated with an increased risk of suffering from symptoms for both women and men, although considerably more pronounced for men, while conflict between demands and control over household work were not. These findings are in line with stress models emphasizing the individual's personal evaluation of the environmental demands in relation to the individual's resources for meeting those demands and a balance is a prerequisite for health.^29,30

The possible conflict between demands from paid work and family obligations as a reason for ill-health, especially among women,¹⁰ was not confirmed in our study even though the women devoting much time to both paid work and household work were at a high risk of symptom reporting. Rather, this study lends support to the beneficial effects of multiple roles and a plausible explanation is that academics, as compared to blue collar workers, often have sufficient job flexibility to reduce work–family conflict and heighten levels of job satisfaction, which will positively affect health outcome.³¹

Methodological considerations

The study participants were matched for age, occupational level and children at home and all the participants were in paid work for at least 35 hours a week. This procedure resulted in fairly homogeneous groups of men and women, which was an advantage in terms of comparing the groups, but limited our results and made them applicable only to white-collar sector men and women. We are therefore unable to draw any conclusions relating to other sectors or to the total population. This also resulted in a restriction of range when it comes to actual working conditions, which possibly explains why only one of the variables (paid work >50 h/week for men) emerged as a statistically significant factor associated with reported symptoms.

As the dependent and the independent variables were based on self-reports from the same instrument, there is a risk of response bias in that the participants who reported a high level of symptoms might rate their working conditions as generally worse than those not suffering from symptoms. However, women suffered more from common symptoms than men, but for all the independent variables the risk of suffering from a high level of symptoms was higher for men, which speaks against such an effect.

The TWL instrument used in this study was developed in Sweden and, when validated, it was found to be a reliable tool for the study of stress-related aspects of the TWL of male and female white-collar employees.²⁰

Conclusion

This study lends support to the idea that men also would benefit from a multiple role situation rather than concentrating mainly on the paid employment role. We found that when care-taking of the children was added to overtime working men's activities, the risk of symptom reporting decreased. Furthermore, men suffered a higher risk of symptom reporting than women when in a situation of excessive overtime work or when perceiving their work burden to be high or when feeling stressed by their paid work. Shouldering multiple roles draws the attention away from paid work and we believe men have something to gain by engaging more in their other roles, i.e. as fathers and husbands. However, more knowledge is needed on the interplay between work and home circumstances, especially on men's roles and health effects.

This project was supported by grants to Ulf Lundberg from the Bank of Sweden Tercentenary Foundation.

References