Work Stress, Obesity and the Risk of Type 2 Diabetes: Gender-Specific Bidirectional Effect in the Whitehall II Study

Setting and population

The Whitehall II study is an occupational cohort established in 1985 with the broad aim of investigating the social gradient in disease outcomes. At study baseline (1985–1988) 10,308 participants aged 35–55 years were recruited from 20 civil service departments in London, UK. After the initial clinical examination, further waves of data collection were carried out in 1989 (phase 2), 1991–1993 (phase 3, including a clinical examination), 1995 (phase 4), 1997–1999 (phase 5, clinical examination), 2001 (phase 6), 2002–2004 (phase 7, clinical examination), 2006 (phase 8) and 2008–2009 (phase 9, clinical examination). The participation rate at phases 3, 5, 7, and 9 (clinical examination phases) was 85%, 71%, 68%, and 66%, respectively. The number of participants in the clinical examination at phase 9 was 6,755. The participants lost to follow-up were more likely to be women and to come from the lower employment grades, were slightly older, had a slightly higher BMI and had a higher prevalence of work stress at baseline.

The current analysis includes 5,138 white participants (3,689 men and 1,449 women) free from diabetes at baseline and with valid data on incident diabetes, psychosocial work stress and all the covariates used in the multivariate analysis (sub-section following). Ethical approval for the study was obtained from the Joint UCL/UCLH Committees on the Ethics of Human Research. All participants gave written informed consent for participation at each phase.

Assessment of glycemia and ascertainment of diabetes

At phase 3 (1991–1993), venous blood samples were taken from fasting individuals (≥8 h of fasting) before undergoing a standard 2-h oral glucose tolerance test (75 g anhydrous glucose over 5 min). Glucose samples were drawn into fluoride monovette tubes which were centrifuged on site within 1 h. Plasma or serum was immediately removed from the monovette tubes, and moved into microtubes and stored at −70 °C. Blood glucose was measured with the glucose oxidase method (15) on YSI model 23A glucose analyzer (mean coefficient of variation at phase 3 2.9–3.3%) (16) and YSI model 2300 STAT PLUS analyzer (phases 5 and 7 mean coefficient of variation 1.4–3.1%) (17) (YSI Corporation, Yellow Springs, OH). Subsequent clinical assessments for diabetes took place at phases 5, 7, and 9. The definition of diabetes used was a 2-h glucose tolerance test finding of at least 200 mg/dl (≥11.1 mmol/l) or a fasting glucose level of ≥126 mg/dl (≥7.0 mmol/l) (18) or physician-diagnosed diabetes and/or use of diabetic medication.

Psychosocial work stress

The Job Strain Questionnaire was developed to provide an integrating theoretical framework for stress-related job characteristics that can be assessed for the full workforce (19). In more detail, the questionnaire assessed the aggregate of psychological stressors affecting work (job demands) and the individual's potential control over job-related decision making (decision latitude).

In the Whitehall study, job demands (4 items; Cronbach's α = 0.67) and decision latitude (15 items; Cronbach's α = 0.84) were measured using the main questions from the Job Strain Questionnaire (19). The empirical association between components of the Job Strain Questionnaire and psychological strain has been previously demonstrated in relation to depression, sleeping problems, and exhaustion (19).

According to the original demands/control model, high job demands were identified as above the median score and low job control as below the median score for the specific sample. Job strain was present when the participant simultaneously scored high on the job demands (above median score) and low on the decision latitude scales (below median score) (19).

BMI and other covariates

Weight was measured by a Soehnle scale to the nearest 0.1 kg with all items of clothing removed except underwear. Height was measured to the nearest mm using a stadiometer with the participant standing completely erect with the head in the Frankfort plane. BMI was calculated as weight (kg) divided by height (m) squared. Obesity was classified according to the World Health Organization definition (20).

Participants reported their Civil Service grade title, which was assigned to 1 of 6 grades based on salary scale. In the British civil service employment grade is an accurate measure of status, income and employment relations and hence socioeconomic position (21). Participants were asked to report how much they were upset from personal illness, death or illness of a close relative or friend, major financial difficulty, divorce, separation or break of a personal intimate relationship, other marital or family problem, experience of a mugging, robbery, accident or similar event. From these questions, a variable scored from 0 to 21 was developed measuring the extent by which participants were upset by life events outside work during the recent past. Other social, psychosocial, and psychological variables (such as marital status, neighborhood deprivation, anger, hostility, social isolation, minor psychiatric morbidity) were available in the Whitehall II study and were considered as potential confounders/mediators but none of these was linked to job strain, thus were not included in the analysis.

Participants reported the frequency of eating a common portion size of each item of a 127-item semi-quantitative food frequency questionnaire. Dietary patterns were identified in sex-specific cluster analysis (PROC FASTCLUS; SAS Institute, Cary, NC). The four clusters identified were: (i) healthy; (ii) Mediterranean-like; (iii) sweet; and (iv) unhealthy (22). Participants reported the number of units of beer, wine or spirits they had consumed in the last 7 days. Units of alcohol (8 g) consumed per week were based on the “Sensible drinking recommendations for adults in the UK” (23) as: no consumption, moderate consumption (1–28 units/week in men; 1–21 in women) and heavy consumption (>28 units/week in men; >21 in women). Frequency and duration of mild, moderate, and vigorous activities were self-reported and hours per week of activity at the three intensity levels was calculated as metabolic equivalent−hours/week. Participants were asked about their smoking status. Participants who reported smoking at phase 3 were defined as current smokers. Those who reported not smoking at phase 3 (and were not identified as current or ex-smokers in previous phases) were classified as never smokers. Ex-smokers were those participants who reported past smoking at phase 3 or current or ex-smoking at previous phases.

Clinical measurements were carried out according to a standard protocol (24). Blood pressure was measured in the sitting position using a Hawksley random zero sphygmomanometer. Venous blood was taken in the fasting state or at least 5 h after a light, fat-free breakfast. Serum triglycerides were measured by automated enzymatic colorimetric methods. High-density lipoprotein cholesterol was measured using phosphotungstate precipitation.

Statistical analysis

Multivariate Cox proportional hazards regression analysis was used to examine the associations between job strain and 18-year incident T2DM. The date of ascertainment of each T2DM case was taken as the mid-point between the date of T2DM identification during data collection (clinical examination or questionnaire) and the date of the previous data collection. Participants were censored at the time of loss to follow-up or at the end of 2004 (phase 7). Type 2 diabetes cases and the censored participants contributed their follow-up time to the overall person-years at risk for the period from 1991 to 2004. Schoenfeld residuals were plotted against follow-up time for testing the proportional hazards assumption (25). All P values were nonsignificant, confirming that the proportional hazards assumption was justified. Hazard ratios (HRs) presented are relative risks for type 2 diabetes comparing participants exposed to job strain to participants not exposed, adjusting for age, socioeconomic position (employment grade), diet pattern, physical activity, alcohol consumption, smoking status, systolic blood pressure, triglycerides, and high-density lipoprotein cholesterol.

The analysis was stratified by BMI, assessing the effect of job strain on incident T2DM among obese (BMI ≥30 kg/m²) and nonobese (BMI <30 kg/m²) participants. The likelihood ratio test was used to statistically test for an interaction between job strain and BMI. The interaction was tested with BMI both as a continuous term (multiplicative interaction) and a binary (i.e., presence of obesity) term. Analysis was performed separately in men and women for investigating gender-specific interactions between stress and obesity.

Summary of findings

In this sample of middle-aged British civil servants, the association between work stress and incident T2DM was modified by BMI differentially among men and women. Work stress was associated with a lower risk of T2DM among nonobese men, while it was associated with a higher risk of T2DM among obese women. There was no evidence for an association between work stress and T2DM among obese men and nonobese women.

The major strength of the current analysis is the accurate assessment of all key factors (psychosocial stress, obesity and diabetes). The Whitehall II study was specifically designed to assess the impact of psychosocial factors on chronic disease and thus assessment of psychosocial work stress is detailed and comprehensive. Diabetes was ascertained by an oral glucose tolerance test at consecutive phases, which is rare for a population-based epidemiological study. Obesity was also accurately assessed using weight and height measured by trained nurses. In addition, the prospective design and long follow-up of the study allows for a detailed investigation of long-term diabetes risk using a large number of incident cases.

Some weaknesses of the current analysis are the self-reported nature of the exposure of interest (psychosocial work stress) and the fact that only a single assessment at one point of time was used here. However, despite being based on subjective data, the job strain measure has been linked to indicators of psychological strain such as depression, sleeping problems, and exhaustion (12). Another weakness is the substantial loss to follow-up, which differed by baseline exposure to work stress and was more apparent for obese women. If however this was a potential source of bias it would have led to a type II error (failing to find an existent association) rather than a type I error (finding an association that dos not exist). In the current results, the magnitude of the effect of work stress among women was relatively big and it may have been even bigger if the obese, stressed women who were lost to follow-up remained in the study.

Potential explanations for the gender-specific stress-obesity interaction

The protective effect among nonobese men. The observed protective effect of work stress on incident T2DM among men is most likely due to prospective weight loss associated with work stress among lean men, as reported in a previous publication from our research group (14). In that analysis, being exposed to work stress was associated with weight loss during a 5-year follow-up among lean men but not lean women, which supports the current finding of lower risk of diabetes among nonobese stressed men but not nonobese stressed women. Even though stress-related excessive weight loss, especially among already lean individuals, is by no means healthy overall, it seems to be protective for T2DM development at least. The reasons for the absence of stress-related weight loss and hence lower risk of T2DM among women in our study need further investigation and this finding is not necessarily applicable to the general population.

The harmful effect among obese women. The main biological candidate for explaining the stress-obesity interaction in relation to T2DM is the stress-hormone cortisol (26). Cortisol can interfere in the normal regulation of blood glucose by altering the body's release and sensitivity to insulin, thus increasing the risk of T2DM (1,2). In a recent publication from the Whitehall study (27), the slope of the diurnal release of cortisol was shallower among obese individuals, highlighting a possible defect in the functioning of the hypothalamic-pituitary-adrenal axis of the stress response. Given that cortisol levels are elevated by exposure to psychosocial stressors (28), a plausible explanation for the observed effect is that stress-related elevations in cortisol levels carry a bigger pathophysiological burden among obese than nonobese individuals.

The observation that obesity modifies the effect of work stress on T2DM only among women could be explained by gender-specific pathways involved in pathogenesis of T2DM. In the Framingham Offspring study, obesity was related differentially to cardiometabolic risk factors among men and women (29). Evidence for sex differences in the development of T2DM, especially in relation to activation of the innate immunity, has been shown consistently in the German MONICA/KORA study (30,31). In addition, there is some evidence for gender-specific psychoneuroendocrine activation with women being more prone to the health impact of chronic psychosocial stress (32,33,34,35). Among a sub-sample from the Whitehall II study, men and women had similar salivary cortisol levels during the weekends but women had significantly higher cortisol levels compared to men in working days (36). Similar results of gender-specific cortisol responses to chronic work stress have been observed in an Italian (37) and a German (38) study. In addition, work stress was linked to decreased heart rate variability (a measure of impaired autonomic activity) among women but not men in a Finnish study (39). In the Whitehall II study, sleep deprivation was linked to higher incidence of hypertension among women but not men (40). This is of importance as sleep deprivation activates the same neuroendocrine pathways as stress and in fact may be a potential mediating factor in the gender differences observed in the current paper. Unfortunately data on sleep duration were not available at the baseline of the current analysis, thus this variable could not be included as a potential mediator.

There may also be a social element in the gender-specific stress-obesity interaction. Obese individuals may be carrying the additional psychosocial burden of discrimination due to their body weight status, making them more vulnerable to other psychosocial stressors (i.e., work stressors). Such discriminations associated with obesity may be felt more strongly among obese women than obese men (41). Exposure to work stress may therefore be an additional burden among obese women. In addition, in the Whitehall II study, women as a whole are more likely to “accumulate” exposure to work stress during follow-up, while the opposite holds for men, who are more likely to “escape” from work stress through the years compared to women. This chronic exposure to stress could provide an explanation for the overall gender differences observed, with higher effects among women.

Conclusion

The protective effect of psychosocial work stress on T2DM risk among nonobese men probably reflects the prospective weight loss related to stress among lean individuals. On the other hand, the harmful effect of psychosocial work stress among obese women probably reflects gender-specific psychoneuroendocrine pathways as well as additional discrimination due to increased body mass. To our knowledge, this is the first study to report the prospective effect of psychosocial work stress on incident T2DM stratifying by BMI. Overall the current results suggest that the stress-obesity-diabetes triangle is complex and gender-specific. We suggest that future investigations on the effect of psychosocial work stress on cardiometabolic disease are performed stratifying by gender and body weight status. Further elucidation of the involvement of psychosocial stress in obesity and diabetes could inform strategies aiming to prevent both diseases.