Introduction
The population mean body mass index (BMI) has increased in recent decades and a prognosis forecasts future increases in obesity prevalence (Breda et al.
2015). While a positive energy balance, originating from excess energy intake relative to energy expenditure, is a fundamental cause for weight gain, psychosocial factors such as mental stress might be contributing factors. Effects of mental stress on weight gain could be mediated through unhealthy behaviors such as low diet quality. In addition, chronic stress could, through higher cortisol levels, impact on visceral fat accumulation and reduction in lean body mass (Kyrou and Tsigos
2009).
Alongside the trend of increasing BMI, the prevalence of self-reported mental stress has also increased (Lissner et al.
2008). Work-related stress, also referred to as job strain, with high demands and low decision latitude as conceptualized by Karasek and Theorell (
1990), has also been reported to be higher in later years (Malard et al.
2015; Utzet et al.
2015) as have the prevalence of other work-related stress exposures (Houdmont et al.
2012). Although changing social norms regarding the stress concept may explain some part of the increased prevalence of self-reported stress, a considerable increase in sickness absence due to psychiatric diagnoses, including the fatigue syndrome, during the last decade has been reported in Sweden, providing support for a factual increase in stress (Swedish Social Insurance Agency
2016).
Job strain has been shown to predict obesity-related diseases such as diabetes (Nyberg et al.
2014), coronary heart disease (Kivimäki et al.
2012), and stroke (Fransson et al.
2015). However, the prospective association between job strain and weight gain, as recently compiled in a systematic review and meta-analysis (Kivimäki et al.
2015), is not consistent and does overall not support the hypothesis that job strain promote weight gain. However, two of the four included studies on job strain in relation with weight gain used self-reported weight, and only one of them was truly population-based, questioning both validity in the outcome and generalizability of the results. In this study, we report on the association between exposure to high job demands and low decision latitude as well as job strain, and major weight gain over 10 and 20 years in a population based cohort of Swedish women and men, aged 30 or 40 years at baseline, with additional focus on the concurrent influence of diet quality.
Results
Baseline characteristics of the study population are shown in Table
1. At baseline, mean BMI was below 25 kg/m
2 in both sexes, but 27% of the women and 39% of the men were overweight or obese. Over the first 10 years, 33.5% of the women and 26.0% of the men gained ≥ 10% body weight, while over 20 years, these figures were 48.9% and 43.7% for women and men, respectively. The mean (sd) weight gain over the first 10 years was 4.6 (6.1) kg for women and 5.1 (6.8) kg for men, while over 20 years, these figures were 6.8 (8.1) kg and 7.1 (7.5) kg, respectively.
Table 1
Baseline characteristics of the VIP study sample
General characteristics |
Aged 29–31 (%) | 36.9 | 36.5 | 37.4 |
Aged 39–41 (%) | 63.1 | 63.5 | 62.6 |
Academic education (%) | 26.3 | 31.1 | 20.7 |
Married/cohabitant (%) | 86.2 | 88.5 | 83.5 |
Current smoker (%) | 24.3 | 26.6 | 21.7 |
Physically inactivea (%) | 47.7 | 48.3 | 47.0 |
Anthropometry |
Height (cm), mean (SD) | 171.9 (9.3) | 165.4 (5.8) | 179.5 (6.4) |
Weight (kg), mean (SD) | 71.5 (12.6) | 64.7 (9.9) | 79.3 (10.6) |
BMI (kg/m2), mean (SD) | 24.1 (3.3) | 23.7 (3.4) | 24.6 (2.9) |
BMI ≥ 25.0 kg/m2 (%) | 32.7 | 27.0 | 39.3 |
Diet |
Energy intake (MJ), mean (SD) | 8.2 (2.7) | 7.3 (2.2) | 9.2 (3.0) |
Sugar (E%), mean (SD) | 7.3 (2.9) | 7.7 (2.8) | 7.0 (3.0) |
Saturated fat (E%), mean (SD) | 14.9 (3.1) | 14.3 (2.8) | 15.6 (3.2) |
Polyunsaturated fat (E%), mean (SD) | 5.2 (1.6) | 4.9 (1.5) | 5.4 (1.6) |
Dietary fibre (g/MJ), mean (SD) | 2.4 (0.6) | 2.6 (0.7) | 2.2 (0.5) |
Fruits and vegetables (g/day), mean (SD) | 226 (166) | 269 (175) | 177 (140) |
Fish (g/week), mean (SD) | 291 (255) | 327 (271) | 250 (228) |
Diet qualityb, mean (SD) | 2.5 (1.3) | 2.7 (1.4) | 2.3 (1.2) |
Age-adjusted analyses in both sexes combined, showed low decision latitude to predict major weight gain over both 10 and 20 years (Table
2), while high job demands were not associated with the outcome. After adjustment for diet quality and other confounders, the effect of low decision latitude remained for major weight gain over 20 years. Interactions between sex and exposure were non-significant (data not shown). Analyses stratified by sex can be found in supplemental Table S1. Baseline diet was itself only associated with major weight gain over 10 years, with an increased risk of major weight gain in women and men reporting a high diet quality [OR (95% CI) 1.26 (1.06–1.50)] as compared to those reporting an intermediate diet quality. When participants with baseline BMI ≥ 25 kg/m
2 were excluded, high diet quality was no longer associated with risk of major weight gain [OR (95% CI) 1.13 (0.91–1.39)].
Table 2
Odds ratio and 95% CI for the association between job demands and decision latitude (mutually adjusted) and weight gain ≥ 10% of baseline body weight over 10 and 20 years in 3872 women and men participating in VIP
High job demandsa | 10 | 1161 (30.0) | 1.07 (0.93–1.24) | 1.14 (0.98–1.32) |
Low decision latitudeb | 1.16 (1.00–1.33)* | 1.12 (0.96–1.31) |
High job demandsa | 20 | 1799 (46.5) | 1.06 (0.93–1.21) | 1.05 (0.92–1.21) |
Low decision latitudeb | 1.29 (1.13–1.47)*** | 1.30 (1.13–1.50)*** |
To study prolonged exposure, 1747 participants (904 women and 843 men) were identified who were classified in the same job strain quadrant category at baseline and at 10-year follow-up, representing 45% of the study sample. Prolonged exposure to low decision latitude predicted major weight gain when women and men were analyzed together (Table
3), while no association was seen for high job demands. However, a significant interaction between sex and job demands was found (
p = 0.02). Analyses stratified by sex showed that prolonged exposure to high demands was associated with an increased risk of major weight gain over 20 years in women, while high job demands did not predict major weight gain in men. A significant interaction was found in women between prolonged exposure to high demands and baseline BMI category (
p = 0.01). Stratification by baseline weight category showed that prolonged exposure to high demands was detrimental only in women with baseline overweight [OR (95% CI) 3.13 (1.67–5.86) vs. 1.24 (0.85–1.79) in women with normal weight at baseline].
Table 3
Odds ratio and 95% CI for the association between prolonged exposure to high job demands and low decision latitude (mutually adjusted) and weight gain ≥ 10% of baseline weight over 20 years in 1754 women and men participating in VIP
Women and men (n = 1747) | High job demandsa | 789 (45.2) | 1.15 (0.94–1.40) | 1.16 (0.94–1.44)c |
Low decision latitudeb | 1.33 (1.09–1.63)* | 1.32 (1.05–1.67)* |
Women (n = 904) | High job demandsa | 424 (46.9) | 1.43 (1.08–1.88)* | 1.54 (1.14–2.07)** |
Low decision latitudeb | 1.23 (0.94–1.62) | 1.16 (0.84–1.62) |
Men (n = 843) | High job demandsa | 365 (43.3) | 0.91 (0.67–1.22) | 0.87 (0.63–1.19) |
Low decision latitudeb | 1.45 (1.08–1.95)* | 1.44 (1.03–2.01)* |
Interactions between job demand and decision latitude were neither significant for baseline exposure nor for prolonged exposure. Results from analyses investigating exposure according to the job strain quadrant model are presented in supplemental Table S2 (baseline exposure) and Table S3 (prolonged exposure) and show that the high strain group was most vulnerable to major weight gain.
Sensitivity analyses (data not shown) of both baseline occupational stress and prolonged occupational stress showed that neither excluding subject who lost ≥ 5% of baseline body weight over 10 (n = 202) and 20 years (n = 224), respectively, nor including participants with incomplete diet data (n = 449) affected the results. Treating baseline exposures as continuous variables did not change the conclusions (data not shown).
Discussion
In this study, we observed that low decision latitude independently and consistently predicts long-term major weight gain in a cohort of Swedish women and men. Moreover, considering prolonged exposure the present study provided suggestive results of a sex-specific effect of high job demands in women, which was more pronounced in women with overweight at baseline.
As far as we know, there is only one previous study investigating associations between the demand-control model and weight gain with comparable follow-up time. That study demonstrated a dose–response association between prolonged iso-strain, defined as high strain in combination with low social support, and incident obesity over 19 years in women and men (Brunner et al.
2007). Furthermore, for the three dimensions of iso-strain—high job demands; low decision latitude; and low social support—significant associations were only found for prolonged low social support at work in men (Brunner et al.
2007). Women did exhibit elevated, although non-significant, odds of obesity for all three dimensions (Brunner et al.
2007). Another study (Niskanen et al.
2017) investigating changes in working conditions in relation with weight gain (≥ 10%) reported that prolonged exposure to high job demands was associated with major weight gain over 12 years in women but not in men, while no association was seen for prolonged exposure to low decision latitude in women or in men. Other previous studies with shorter follow up of between 2 and 7 years have not found evidence of an association between baseline job strain and weight gain or risk of obesity in either sex (Ishizaki et al.
2008; Nyberg et al.
2012), or only found associations in women (Kivimäki et al.
2006; Roos et al.
2013; Shields
1999). In models adjusted for several covariates, Shields (
1999) found females with high baseline job strain to have increased odds of a weight gain of more than 8.8% of baseline body weight over 2–3 years, while Roos et al. (
2013) showed that females belonging to the passive group had an increased odds of gaining 5 kg or more over 5–7 years. Furthermore, Kivimäki (
2006) found a positive association between job demands and weight gain in women. Comparability of study results is problematic because of discrepancies in both exposure and outcome assessment. Moreover, although most previous studies reported analyses stratified by gender, support for gender specific effects of job strain in relation with weight gain was inconsistent. However, in accordance with results reported by Niskanen et al. (
2017) and supported by a significant interaction between sex and high job demands, we found prolonged exposure to high job demands to stand out as detrimental only in women. One plausible explanation is that a female sensitivity to prolonged high job demands could relate to women’s double work load due to higher responsibility for domestic work, which appears also in a relatively gender-equal society as Sweden (Harryson et al.
2012).
The link between job strain and weight gain could potentially work through, for example, dietary intake, physical activity, or metabolic rate, but neither physical activity nor diet quality explained the associations seen in the present study. The diet quality index used in the present study, which was developed in another Swedish cohort (Drake et al.
2011) has previously been shown to inversely associate with total mortality as well as cardiovascular disease morbidity and mortality (Drake et al.
2013; Hlebowicz et al.
2013), while no association was found for diabetes risk (Mandalazi et al.
2016). The diet index has to our knowledge not previously been used in prediction of weight gain, and it is plausible that such an index is not optimal for that purpose, due to obesity-related reporting biases. Because neither changes in metabolic rate, nor changes in body composition or regional fat pattering were measured in the VIP cohort, we can only speculate in that exposure to job strain, with potential exposure to higher levels of the catabolic hormone cortisol, could have had detrimental effects on muscle mass with lower basal metabolic rate and weight gain as a consequence. The primary strength of this study is the repeated assessment of both exposure and outcome with the ability to investigate associations between prolonged exposure to high demands and low decision latitude, respectively, and weight gain over one and two decades. The examination interval of 10 years was not explicitly chosen for the present study, but was likely relevant for our outcome of interest, since many previous studies with shorter study interval have failed to show significant associations between job strain and major weight gain. The use of validated instruments to assess exposure to both job strain and other lifestyle factors, and standardized measurement of outcome performed by medical staff also deserve to be highlighted as important strengths. The outcome was defined as a relative gain in baseline weight of 10% or more, and an alternative would have been to use an absolute cut-off for weight gain instead such as that of 5 kg or more as suggested by World Health Organization (
2003). The use of a relative weight change cutoff is, however, supported by increased congruence across body sizes (Stevens et al.
2006).
This study has potential limitations that need to be considered. First, it has to be acknowledged that the exposure to the two dimensions of job strain is self-reported by participants and may introduce various forms of bias to the analysis and results. Possible fluctuations in exposure during follow-up must also be acknowledged, although such fluctuations would be likely to attenuate our observed effect estimates. Second, although the full VIP cohort is population based with an essentially representative participation (Weinehall et al.
1998), the current VIP sub-sample that attended three subsequent examinations, may be subject to participation bias. However, we recently showed that a sub-sample of the VIP cohort with two repeated measures (Winkvist et al.
2017) did not exhibit any essential socio-demographic differences compared to the full VIP cohort. Third, we were only able to use baseline data for dietary intake, since the assessment method was changed from an 84 item FFQ to a 64 item FFQ at the follow-ups. Finally, the present study lacked data on metabolic rate, body composition, and regional fat patterning.
In conclusion, the present study supports the hypothesis that occupational stress promotes weight gain. Neither baseline diet quality, physical inactivity, nor other covariates, explained the results, therefore, pointing at other mechanisms underlying these associations. To further understand the mechanisms between job strain and weight gain, studies on changes in both regional fat patterning and body composition in relation with job strain are warranted. Although the association between occupational stress and major weight gain seems rather weak, the public health relevance of our findings could be important because of the increasing prevalence of work-related stress. The importance of these findings is likely to go beyond the risk of unhealthy weight gain, since the identification of vulnerable groups and gender-sensitive preventive efforts to reduce job strain have the potential to not only reduce weight gain, but also risk of cardiovascular disease and diabetes.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.