Psychosocial and other working conditions in relation to body mass index in a representative sample of Australian workers

The prevalence of overweight and obesity has been high in most industrialized nations, since the early 1950s; however, this trend accelerated in the 1990s. Australia has not escaped this phenomenon [1]. The risk of cardiovascular disease increases with overweight and obesity [2]. According to the Australian National Health Survey conducted in 2001, 16% of men and 17% of women (aged more than 18) were obese, 42% of men and 25% of women were overweight, and 1% of men and 5% of women were underweight [1].

An extensive literature has been published on job stress and cardiovascular disease (CVD), mainly among men [5]. This literature demonstrates that job strain and high effort/low reward conditions predict CVD [6–8], but the relative contributions of direct and indirect mechanisms remain unclear. Some evidence suggests indirect effects of psychosocial and other work conditions on health through health behaviours [5].

Indirect pathways may include effects of job stress on physical activity, eating behaviours, and other behaviours that may be related to BMI [9]. Previous studies have found associations between working conditions and health behaviours, such as diet, physical activity and alcohol consumption, which impact weight change [10–12]. For example, in a cross-sectional analysis of Japanese workers (n = 6,759) job strain was associated with low vegetable and high alcohol consumption [11]. Another cross-sectional survey of American workers (n = 3,843) showed that job demands were positively associated with high fat intake in men, while decision latitude was positively associated with physical activity in both men and women [12]. More recently, the Helsinki Health Study (n = 6,243) found that among women, mentally strenuous work and high job control were associated with a healthy diet [10].

The two most widely used instruments to measure occupational stress are Karasek's demand/control (DC) and Siegrist's effort/reward imbalance (ERI) models. The demand/control model focuses on task-level job characteristics. It postulates that psychological strain results from the interaction of job demands and job control, with the combination of low control and high demands producing "job strain". In contrast, the effort/reward imbalance model includes personal characteristics of the worker (i.e., over commitment) and also conceptualizes and measures work conditions more broadly than the demand/control model. It focuses on the reciprocity of exchange at work where high costs/low gain conditions (i.e., high effort and low reward) are considered particularly stressful.

While the demand/control and effort/reward imbalance models have been tested in relation to CVD outcomes they have been less widely investigated in relation to risk factors for CVD including body mass index (BMI) [12]. Nonetheless, fourteen studies have been conducted using job stress to test for associations with body mass index [4, 13–25]. The DC model was used in ten of these studies [12–16, 20, 21, 23–25], two of which also used the ERI model [13, 14].

Six of the studies with the DC model showed a positive and statistically significant relationship with BMI [12–15, 20, 24] but the remaining four showed no association. Both studies with the ERI model showed a positive and statistically significant relationship with BMI [13, 14]. And, two of the remaining four studies utilizing other measures of job strain showed positive and statistically significant relationships with BMI [4, 18].

In these positive investigations with the DC and ERI models, high job strain [13], low control [13, 24] and high ERI [13] have been linked with increased BMI. Among the four workplace studies of stress and obesity that did not use the DC or ERI models, the results varied [4, 22, 26, 27]. House et al. [22], using data from the Tecumseh Community Health Study from 1967 to 1969, demonstrated negative associations between occupational position and "pressures on the job" in relation to obesity for both men and women.

In a study of 1,137 Swedish women, Rosmond and Bjorntop [4] found that education, satisfaction with management, attempts to alter work situation" (i.e., a proxy measure of participation and control), and physical exercise were all negatively associated with BMI.

In a cohort study, Kornitzer and Kittel [26] found no association between job stress and obesity. The stress measure they used in this study of obesity was also tested in relation to coronary heart disease and was also non-predictive. Conversely, Georges et al. [27] found a borderline significant positive association (P= 0.06) between job demands and increased BMI. This survey also demonstrated that men with high job strain were more likely to have a pattern of central body fat distribution then men with low job strain [27].

Non-psychosocial working conditions have been investigated in relation to body mass index also. Using data from the National Population Health Survey in Canada, Shields [28] demonstrated that, after statistical adjustment, men who worked more than 35 hours a week had an odds ratio of 1.4 for being overweight (BMI> 25). No association between long hours of work and overweight was demonstrated for women.

Using a representative survey of workers in the Australian state of Victoria, we have assessed associations of job stress (i.e., demand/control; effort-reward imbalance models), shift work, physical demand and hours worked, with BMI among men and women.

This study was reviewed and approved by the University of Melbourne's Human Research Ethics Committee (HREC protocol #030398).

Model 4 is the most parsimonious final model presenting significant work-related predictors of BMI in men for our sample. The unstandardised regression coefficients show that men with high physical demand have a mean BMI that is 1.04 BMI units less than those in low physical demand jobs. In contrast, men with longer working hours have a mean BMI that 1.37 units higher than those working up to full time, and increasing in stepwise fashion, those working excessive hours have a mean BMI that is 1.88 units higher than those working up to full time.

For women, there were no statistically significant associations between working conditions and BMI after including DCM and ERI, independently or in combination, along with socio-demographic adjustments (data not shown).

There are four main results which arise from this study. First, while the effort/reward imbalance ratio itself is not associated with BMI, high effort in men and low reward in women were associated with BMI. Second, for men, high psychological demand, as measured in the DC model, was positively associated with BMI. Third, after controlling for main job stress measures (ERI or DC), for men, longer and excessive working hours were positively – and physical demand was negatively – associated with BMI. Fourth, among women, after similar adjustments neither psychosocial or other working conditions were significantly associated with BMI.

Our population-based sample is representative of occupational groups in the general population. Therefore, our findings can be generalised to the general working population in Victoria. Further, this study was strengthened by the comparative assessment of a range of psychosocial and other working conditions (two measures of job stress, working hours, physical demand, and shift work).

However, three limitations must be taken into account. First, the cross-sectional design of our study cannot support causal inferences between occupational factors and BMI. Second, information on independent and dependant variables were collected using self-reports. Even if the questionnaire was designed to minimize self-report bias in responses, some items may be subject to this type of bias. In particular, psychological demand has been shown to have a strong subjective component. And, self-reports of BMI produce under-estimates as men tend to overestimate their height and women tend to under-estimate their weight. Use of self-reports for BMI, as in this paper, will tend to attenuate any associations observed.

Studies of psychosocial working conditions and BMI have demonstrated an association between high job strain [13], low control [13, 24], and high ERI [13], "pressures on the job" [22], a proxy measure of participation and control [4], and a borderline association between high psychological demand and BMI [27]. These results indicate that demands, control, or some combination of these may be associated with increased BMI but clear results from this small number, of largely cross-sectional studies, do not emerge.

The results from our study suggest that "demand" rather than "control" factors may be more salient in relation to BMI, at least for men. After fully controlling for confounding, two workload-related factors, psychological demand (from the DC model) and working long hours were both strongly associated with BMI. As well, in univariate models high effort was associated with BMI. Given that high effort combines measures of physical and psychological demand into a single variable, and given that high physical demand is negatively associated with BMI and high psychological demand with increased BMI, the effects of one may cancel the effects of the other, at least in relation to BMI.

There are plausible mechanisms through which long working hours could be related to BMI. Working long or excessive hours may reduce the opportunity for leisure time physical activity, might also increase the frequency of eating higher caloric value take-away or restaurant food, and might also lead to higher alcohol consumption as an unwinding or coping mechanism. It is important to note also that this assocaiton is of a similar magnitude to the effect of increasing age on men (data not shown, highest age group [51+ years)] has mean BMI 1.60 units higher than youngest [<30 years]).

In this study, while half the men worked longer working hours and one third excessive working hours less than half the women worked up to full time hours, 38% worked longer hours, and only 15% worked excessive hours (Table 1). Thus, working more than 35 hours per week is mainly a male phenomenon so that the lack of observed association between working long hours and BMI among women may be due to the small numbers of women working long hours.

Our findings suggest that psychosocial work conditions may impact BMI, particularly among men, and that largely independently of job stress, both low physical demand at work and longer working hours among men may increase BMI.