The present study indicated that the risk of excessive fat accumulation in adipose tissue might be a consequence of exposure to several occupational or work-related factors and differs markedly in male and female workers. Previous studies observed sex differences in BMI levels and obesity prevalence in the same occupational group or with reference to specific occupational variables (20, 30, 31). Although these data have been explained by calling into question demographic, socio-economic, and cultural or lifestyle factors, the common conclusion was that further research was needed on this topic. We obviously cannot exclude the possibility that our results related to sex differences may be partly due to determinants other than those we examined. However, the observation that in the adjusted analysis male and female workers showed different susceptibility to obesity, even after taking account of some key variables (age, education, sedentary work, psychosocial stressors and chronic diseases) suggests that, at least for some occupational factors, sex-specific differences have a significant role.
Therefore, assuming that this hypothesis is valid and considering workplaces as optimal and privileged adult settings where it is possible to plan and implement interventions to prevent and treat obesity, in our opinion these measures should certainly include - though not necessarily be limited to – promotion of healthy lifestyles. Regular physical activity and/or a healthy and balanced diet are general recommendations that are always valid for all people (or workers) regardless of sex, demographic and socio-cultural variables and type or characteristics of their work. Therefore, if we really want to take advantage of the workplace to work out more strategies to win the battle against obesity, we have to begin by considering that this condition is not just a consequence of a personal choice, but is caused more by a complex interplay between an individual and his or her environment (14). To properly address this issue we need more qualitative and quantitative data to answer unsolved questions. For example, what are the occupational groups or worker categories that are at greater risk of obesity? Are there other occupational risk factors besides sedentary work and diet that may be associated with obesity? Are they modifiable? Considering the increase of women in the labor force and their growing involvement in roles and activities that were traditionally male-focused, might there be some gender differences in work-related factors potentially associated with obesity?
Our findings provide helpful information to tackle some aspects of these questions. First of all, analysing the main socio-demographic characteristics of the sample (sex, age and education), higher BMI and prevalence of overweight or obesity were observed in males, at older ages and in workers with lower education (Table 1). These results are in accordance with previous evidence, confirming that these variables are correlated with obesity (20-22, 30, 31, 38). Therefore these socio-demographic determinants should be taken into account when trying to establish correlations between BMI and occupational groups or work-related factors, since they may help explain some differences in BMI (31).
The analysis of BMI in different occupational sectors indicated that male workers involved in healthcare and social assistance had the highest prevalence of overweight and obesity, while their female counterparts had higher prevalences in agriculture, fishing and hunting (overweight) and in construction, healthcare and social assistance (obesity) (Table 2). These data are in line with previous studies, further confirming that occupations requiring sedentary behavior or low levels of physical activity are marked by higher prevalences of overweight and obesity (although the association was weaker in women) (21, 30, 31, 38). Nevertheless, there are some important differences since, differently from our results, several other groups reported higher prevalence rates in transportation and warehousing workers (20, 30-32, 38) and lower levels in the healthcare sector (32, 38). In our study, using the ATECO classification, transportation and warehousing workers were included in the same group as information and communication workers, while in other studies workers in the healthcare and social assistance sector, with different jobs (e.g. healthcare practitioners and technical, healthcare support and protective services), were considered separately (32, 38). Therefore, it is quite likely that these conflicting results are due to differences in classification of some occupational sectors although we cannot rule out an influence of socio-demographic and cultural factors (31).
With regard to socio-demographic characteristics, we also conducted an adjusted analysis to explore to what extent the BMI (and related overweight/obesity prevalence rates) in the occupational groups were affected by differences both in the distribution of these variables and other work-related factors (Table 2). The findings confirmed that, for female workers, these determinants could have a significant impact, changing the OR in specific occupational groups. On the other hand, few occupational categories (agriculture, fishing and hunting and other public and personal services) were associated significantly with increased OR even after adjustment, thus suggesting that other work-related factors, specific for these occupational groups and not captured or assessed in the present study, can contribute to overweight and obesity.
In addition, the differences in occupational effects by gender indicate that sex-specific factors other than socio-demographic and work-related determinants may influence the likelihood of overweight and obesity. Therefore additional research investigating the reasons for weight disparity between male and female workers in different job categories is much needed (20, 31).
Besides knowing the obesity prevalence rates for the different worker categories, it is also important to understand the reasons for them, in other words to identify the work-related factors (besides sedentary work, jobs with low physical demand, nutrition at work) that could possibly be associated with obesity. Subjects who worked long hours had a high risk of overweight/obesity (35, 40); similarly, working ≥35 and >40 or >50 hours per week was significantly associated with increased BMI in men (41) and with obesity in workers of both sexes (22, 38), whereas Kim et al. (36) found an association between this condition and long working hours only in women workers. We found no significant association between long working hours and overweight or obesity, even though women who worked >40 and ≥55 hours/week had the highest prevalence of overweight and obesity, respectively. Surprisingly, the highest level of obesity was seen among the men with the lowest working hours.
Shift workers had higher prevalence rates for overweight and obesity and the difference was significant for females. Luckhaupt et al. (38) and Di Milia and Mummery (35) obtained similar results, reporting higher prevalences of obesity and BMI levels in shift workers on night or rotating shifts compared to those on day or evening shifts. The increased odds of overweight/obese in female workers persisted after adjustment for socio-demographic characteristics, variables related to health and safety protection and chronic diseases (Table 4) and the trend was similar when taking night shifts (1 or 2 times per week) into account (Table 5). However, male shift workers were associated with an increased OR only when considering age, education or health surveillance and there was no significant association with night shift work (Tables 4 and 5).
These findings suggest that the relations among shift, night work and obesity is influenced by gender-specific variables. It has been suggested that inadequate or difficult working conditions can trigger a stress response that in turn may enhance the risk of obesity (42). Indeed, when a person experiences a stressful condition the production of hormonal factors (especially of adipokines, which are strongly linked to appetite and fat storage) changes substantially (43-45). Since the key sex differences in fat storage in men and women include different insulin sensitivity and adipokine production it is plausible to hypothesize that the gender-specific differences observed in our study are due to this sex asymmetry (13, 18, 19).
Then, assuming that stressors play a role in weight gain, few studies have investigated the association between certain psychosocial working conditions and obesity (21, 38). In this regard, a correlation has been reported between hostile work environment and, to a lesser extent, job insecurity (38), whereas in the study by Choi et al. (21) job demand, supervisor and/or co-worker support were not associated with increased obesity prevalence and only low job control in female workers showed a significant difference. Our data are similar to those of Choi et al. (21) as no association was established between male and female workers’ exposure to work-related stress and prevalence rates for overweight and obesity.
We also studied the type of contract (permanent or temporary) as a possible work-related stress factor since a temporary contract is a major source of concern about becoming unemployed. This variable was taken into consideration by Luckhaupt et al. (38) who, while noting a higher prevalence of obesity among temporary workers compared to permanent ones, failed to identify any significant relation between boundary work and obesity. Similarly, the present study found no overall differences as regards the type of contract. However, the logistic regression models provided interesting findings highlighting important differences between male and female workers (Table 3). For women permanent contract seemed to serve as a protective factor against the risk of obesity, whereas this variable was associated with a higher OR in men.
Explaining these conflicting results is challenging since several other social, cultural and work-related factors (that were not analysed here) could be responsible for the differences. Nevertheless, once again, the divergent results for the two sexes suggests the need to focus primarily on the role of biological/physiological sex differences in facilitating or combating obesity in the workplace. Further research should verify whether exposure to specific occupational risk factors (long working hours, shifts and night work, psychosocial stressors) can influence - and how - the expression of these biological and physiological characteristics, or the functioning of some organ systems (e.g. the endocrine system) that could therefore explain the different propensities to obesity in male and female workers, because of the significant impact on metabolism and adipose tissue storage.
Finally, our data showed a significant association between overweight and obesity prevalence with several chronic conditions such as musculoskeletal, respiratory and cardiovascular diseases, both in males and females (Table 1). These results further underline the importance of preventing and treating obesity as excess weight gain is an important risk factor for several non-communicable diseases.
Strengths, limitations and future perspectives
The study has some strengths. First, it addresses a large representative sample of the Italian working population, filling the gap of the lack of studies on occupational factors and risk of obesity in this country. Studies on national representative samples add value when investigating overweight and obesity, since there is ample evidence of external socio-cultural factors, such as diet, culture, acceptable lifestyles, and behavioral patterns affecting a person’s weight. Secondly, we included several occupational factors in the risk of obesity by adjusting for main confounders. Some studies have looked at occupational aspects, but they mostly refer to specific occupational populations and sectors. Moreover, we considered a large set of occupational variables including some that are generally less investigated (e.g. work shifts, night work, type of contract) in relation to obesity, and relevant from the perspective of gender differences. Finally, this study is not limited to checking for gender, but takes into consideration gender-specific differences in studying occupational factors associated with the risk of obesity in order to characterize the risk of obesity categories better and help identify workplace-targeted intervention strategies.
Some limitations must be addressed too with a view to future improvements. First, the cross-sectional design allows us to describe associations but not causation. In other words, we cannot draw causal inferences about the effects of the different variables on overweight and obesity since we cannot define the direction of the associations. However, data were collected as part of a national project, INSuLa, a well-established worker population survey, on a representative sample and included reliable information on several socio-demographic variables and working conditions.
This survey is now becoming a monitoring system to follow changes over time. The data collected in the study also gave some suggestions on how to integrate useful measures in the next waves attitudes and behaviors related to meals at work and physical activity will be considered in the future and linked to the BMI.
As second limitation is related to the self-reporting of body weight and height. Consequently BMI calculations are subject to error and our findings might be vulnerable to reporting bias. People have a tendency to overestimate their height and underestimate their weight and this self-reporting bias is stronger among overweight and obese individuals (46, 47). However, in adults measured and perceived BMI are strongly correlated (48) and a limited number of studies analysing the differences between self-reported and measured anthropometrics in selected working categories have provided evidence that self-reported weight and height information is a reliable tool to assess BMI in large worker samples (49-51). Future studies might investigate the validity of self-reported as opposed to measured BMI in specific Italian occupational groups, considering some sociodemographic differences such as sex, age, and education).
Finally, the study is based on a standardised questionnaire to survey Italian workers on their perception of health and safety at work. Even though we examined a broad set of factors related to work that can affect the BMI, some factors such as life satisfaction, job satisfaction, or income could not be investigated. These factors were not present in the standardised questionnaire aiming to provide a broad overview of different factors related to work and health, but might usefully be considered in the future in the light of our findings.