This cross-sectional study based on the baseline assessment of employees participating in a group-randomized trial for weight gain prevention suggests that high job strain, as a construct for chronic stressors in the work environment, and not job insecurity, as an acute stressor, is associated with weight status. Furthermore, there is no synergistic effect of chronic and acute stressors on the outcome. In this sample of employees it appears that high job strain has an effect on weight status through the adoption of unhealthy behaviors such as not engaging in leisure time physical activity and practicing sedentary behaviors (e.g., watching TV) since the effect of job strain all but disappears once the behavioral variables are added to the models. Diet quality, as operationalized by the intake of fruit and vegetables, does not seem to mediate the effect of a high job strain environment on weight status. These results are fairly consistent regardless of the way weight status was modeled. Unexpectedly, employees in an active work environment are more likely to be overweight or obese and its effect does not seem to be mediated by any of the postulated intermediate variables.
A fair comparison of the results of the current study with previous literature is difficult since there are differences in the way the psychosocial work environment is defined, the confounders included in the analyses, and the measurement of weight status (BMI, overweight, weight gain). In a recent review9
of 18 investigations examining the components of the demand-control model and body weight published between 1989 and 2006, 15 examined the effect of high job strain, of which 12 studies were cross-sectional and 3 were prospective. According to this review, the evidence is inconsistent since in 3 of the studies the association between high job strain and a measure of body weight was fully confirmed, in one was partially confirmed, and in 10 there was no association. In a subsequent comparative study of pooled cross sectional data from three worksite cohorts in Britain, Finland, and Japan, the high strain, passive, and active quadrants of the demand-control model were not associated with obesity36
when compared with the low strain quadrant neither in the unadjusted nor in the fully adjusted models. On the contrary, in the prospective Whitehall II study,37
repeated measures of work stress over a period of 19 years showed a dose response relationship with incident overall obesity (BMI >=30kg/m2
) and central obesity once adjusted for a variety of health behaviors (alcohol consumption, smoking, daily intake of fruit and vegetables, fiber consumption, exercise). Work stress, however, is defined as high job strain and lack of work social support and, thus, it is not fully comparable to the effect of job strain alone as examined in the study presented here. The inconsistency of the relationship between job strain and body weight might be explained by the findings of two prospective studies included in the previous review.9
These two studies among British civil servants38
and Danish men22
indicated that the longitudinal effects of job strain on weight differ according to initial BMI. Whereas high job strain predicted weight gain among overweight or obese employees at baseline, high job strain predicted weight loss among the leanest ones. Other potential explanations for the conflicting results is that different population of workers may have diverging coping strategies allowing them to handle high job strain in different ways. For example, age, social support in and outside the workplace, work-home interface,39
and the existence of a social safety net may influence the way workers respond to a stressful work environment. In addition, different individual responses to ‘stress’ through eating and physical activity have been reported in the literature26,40-41
and may explain the lack of consistency across populations and study designs.
Our study adds more evidence of a lack of cross-sectional association between high job strain and body weight. Although the magnitude of the association was relatively strong after adjusting for confounders, measures of leisure time physical activity and sedentary behavior remove the effect of job strain on weight status. High job strain was associated with low leisure time physical activity among men in London and women in Helsinki14,36
but was not related to measures of exercise among workers in Minnesota.42
To the best of our knowledge, there are no studies of the effect of high job strain on TV viewing as a measure of sedentary behavior. There is, however, evidence of an association of hours of TV viewing with weight gain31
and overweight and obesity35
among adults. Thus, further investigations of sedentary behaviors as measured by TV-viewing and high job strain are needed to confirm the mediating role of the latter behavior with weight status.
We hypothesized that physical activity, sedentary behaviors, and the number of servings of fruit and vegetables per day as a proxy for diet quality would be one of the mechanisms for which high job strain would have an effect on measures of body weight. The hypotheses were partially confirmed since our measures of physical activity and sedentary behaviors were associated with weight status in our sample and modify the magnitude of the association between job strain and weight status. In contrast, this study did not confirm our hypothesis regarding diet quality. It is possible that the fruit and vegetable FFQ used was not an appropriate measure of diet quality since it does not include other components of a healthy diet such as whole grain breads and cereals and the quality of the fat intake (saturated versus unsaturated fat). Although several studies have found no relationship between high job strain and diet quality measured as the consumption of fruit and vegetables,23,43-44
one Japanese cross-sectional study found that high job strain ratio was related to a higher intake of fat among men.15
Again, the varied tools used to measure diet quality preclude an accurate comparison of the findings. Interestingly, Jeffery and French31
found that energy intake and percentage of energy from fat were also positively associated with TV viewing. It does raise the possibility that in our study, TV viewing might have inadvertently served as a proxy for caloric intake and diet quality and may have counteracted the effect of our proxy for diet quality. Finally, it is important to note that characteristics of the psychosocial work environment have been hypothesized to have a direct effect on health outcomes including overweight and obesity independently of their effect on individual behavior.9
An unexpected finding was that active work environment was associated with overweight or obesity. Among Minnesota workers, active job environment was not associated with BMI.42
Since the effect of an active job environment was only observed in the adjusted analysis and was not mediated by other variables, we suspected that the relationship might be confounded by age since employees in an active work situation have more resources to cope with high psychological demands and the latter may be related to seniority. However, in our data there is no difference in mean age between workers in high job strain situations and all other quadrants included active environments (data not shown). In synthesis, further investigations of the characteristics of employees who work in active job environments in our sample will help understand this association.
Given that at the time of baseline assessments in the current study, the company was implementing massive lay-offs, we hypothesized that job insecurity would work as an acute stressor in the workplace and would act directly or indirectly on measures of weight status.22
Also, we expected to find an interaction between acute and chronic stressors. In our sample, we were unable to confirm our hypotheses. Health effects on lay-off survivors are generally understudied.20-21
It is plausible that our measurements were done too soon to be able to capture any effects on employees’ weights.
This study has several limitations. One of which is the impossibility of providing causal inferences given the cross-sectional nature of the data. Also, we do not know how representative our sample is of our worksite population since the company lacks information for relevant comparisons. Finally, the 3 item subscale assessing job insecurity as an acute stressor does not necessarily represent current job insecurity. One item of the subscale represents job insecurity in general, another represents current job insecurity, and the third represents future job insecurity. We expected, nonetheless, that in our particular context this subscale would capture the current situation. Although employees’ concern about their job security might have existed for a long time, the situation at the moment of baseline data collection went beyond what the company had previously experienced as evidenced by the unprecedented magnitude of the layoffs and the fact that entire worksites and their operations were decommissioned. The strengths of the study are that we count with measured height and weight and self-reported instead of imputed job strain measures. These data suggest that wellness programs in the worksite should target health enhancing behaviors to minimize the health effects of psychosocial work conditions. At the same time, worksites should examine their organizational and personnel development in order to prevent stressful work environments. For example, future research should investigate mechanisms that could reduce employees’ demands or increase their control over their jobs such as supportive supervision and changes in the structure of jobs. Further studies on the effect of the psychosocial work environment on weight status should be prospective, examine the potential bidirectionality of the effect of job stress by stratifying for baseline weight status, include measures of individual psychological features and work-home interface that can moderate the effect of working conditions. More evidence is needed on the health of lay-off survivors, in particular under the current economic downturn.