Poor nutrition and inadequate physical activity (PA) are lifestyle behaviors resulting in obesity and a host of chronic diseases [1–4]. The scale of the public health problem necessitates that efforts to promote healthy weight reach large portions of the population; however, existing efforts have had limited success [5, 6], as many have focused on individuals and have targeted behavior change through psychosocial and cognitive–behavioral strategies. Ecologic frameworks highlight the importance of factors beyond the individual level and show how the individuals’ environments can be related to their behaviors [7, 8]. Theories of environmental and health interactions have also suggested that an environment may promote or prevent healthy behaviors through the physical, interpersonal, organizational, and sociocultural characteristics of a setting [8].

Considering the potential for environments to impact health beyond the individual level, worksites may be an effective setting for efforts to promote healthy weight; according to the American Time Use Survey, on average, adults spend 8.8 hours per day in work and work-related activities (http://www.bls.gov/tus/charts/). Researchers have found relationships between workplace supports (e.g., incentives and facilities that support active transportation such as bikelock areas and showers or access to Employee Assistance Programs) and employee health and well-being [9, 10] and health behaviors [11, 12]. In addition to specific workplace supports, the social environment in the workplace can influence employee health. The social environment can influence behavior by “shaping norms, enforcing patterns of social control (which can be health promoting or health damaging); providing or denying opportunities to engage in particular behaviors; and reducing or producing stress,” as well as by “placing constraints on individual choice” [13]. Behavior theories suggest the importance of social norms, including those in the worksite, in determining health behaviors [14–16], and norms have been shown to influence obesity [16] and other health-related behaviors [17]. Surrounding culture and social environment can influence obesity and behavior [15, 18], and the workplace environment is particularly important [19]. Social norms in the workplace, in particular, may influence workplace outcomes such as safety behaviors [20]. However, while studies exploring correlates and interventions in the environmental realm have shown benefits on diet and activity behaviors, research on social environment changes has been more limited [11, 21, 22]. Specifically, the literature showing empirical relationships and precise leverage points for influencing employee health is sparse. One study has investigated the relationship between such social/organizational characteristics as worksite support and co-worker behaviors, however, this was limited to hospital employees in one hospital system [11]. Additionally, previous research has shown that demographic characteristics of workers (e.g., race and gender) and workplace size are associated with diet and activity behaviors [23–29] and/or workplace characteristics [30, 31]. Therefore, these factors need further study. This paper explores the relationship between workplace social environment and cultural factors and diet and PA behaviors and obesity in a large sample of employees from a diverse set of workplace settings across multiple metropolitan areas.

The social/organizational environment variables were dichotomized (strongly agree and agree vs disagree/strongly disagree). Diet behaviors included eating fruits (at least 1 time per day), eating vegetables (at least 1 time per day), eating snack foods, and eating fast food (at least two times per week). These variables were dichotomized based on the low prevalence of fruit and vegetable intake among U.S. adults [45, 46], the contribution of fast-food to U.S. diets [47], as well as the distribution in the data. PA behaviors included reporting at least 150 minutes of PA per week [42] and obesity (as assessed by BMI) was dichotomized at 30 (with BMI≥30 being the definition of obesity) [37]. Bivariate associations were explored between the social/organizational environment, diet and PA behaviors, obesity, and demographics. Logistic regression models explored the association between social/organizational environment variables and diet, PA, and obesity; these were conducted with and without adjustment for demographic factors (race, employer size, age, sex, and income). Variables for adjustment were selected based on the associations in the bivariate analyses and those common in the nutrition epidemiology literature. To be consistent, we included the same adjustment variables in all analyses.

There were important differences between reports of social/organizational characteristics by those with different demographic characteristics. Our exploratory study suggests that reports of seeing co-workers doing several types of healthy behaviors (eating fruits and vegetables, doing PA, and doing PA on breaks at work) seemed to be related to a number of outcomes including fruit and vegetable intake and PA, and these all increased as worksite size increased. It is possible that larger work places offer more resources and greater numbers and diversity of co-workers, thus allowing employees more opportunities to observe their co-workers engaging in healthy behaviors. This fits with behavior theories indicating the importance of norms [14–16]. There were additional differences by gender, race, and income in the reporting of these characteristics. Further, we found associations between agreement with the statement that the company values the employee’s health and important behavior/obesity outcomes, indicating employees agreeing with the statement report obesity-preventive behaviors, though these were somewhat reduced with adjustment. It is possible that these companies provide more resources toward employee health and well-being. As with seeing co-workers engaging in healthy behaviors, report of agreement with this statement that the company valued the participant’s health also varied by demographic characteristics, for example, reports of agreement were less frequent among Blacks, those reporting lower income, and younger workers. This suggests a culture of healthy behaviors in a workplace is related to worker health behaviors [48].

Others have found positive relationships between normative eating and activity behaviors (i.e., perceived behaviors of co-workers) and dietary and PA behaviors [11, 22, 49]. Further, among hospital employees, Lemon et al. [11] found relationships between organizational commitment to employee health and BMI and Sliter et al. [49] found that the workplace climate for healthy weight maintenance was uniquely predictive of most individual variables, diet quality and PA, in particular. However, these results have not been consistent in the literature. Basen-Engquist et al. [50] did not find relationships between organizational climate and employee behaviors. Our results indicated stronger relationships between normative health behaviors and dietary behaviors than those for PA. Lemon et al. found similar results and suggested this may be a result of the routine activity of eating, but not exercising, as a part of the workday [11].

The current study found important differences in perceptions of workplace support based on demographic characteristics; for example women were more likely to report seeing co-workers eating fruits and vegetables than men, and White, older, and higher income workers were more likely to feel the company values health than younger, Black/African American or other, or lower income employees. The discrepancy by gender has been seen in previous studies [11, 22]. In a study of hospital employees, women had higher perceptions of coworker healthy eating behaviors than men, and Hispanics and Asians/others had higher perceptions than non-Hispanic Whites. Further, non-Hispanic Blacks had higher perceived organizational commitment to employee health than non-Hispanic whites, which is the opposite of what we found in the current sample. Other studies have not found differences based on ethnicity [22].

This study has several implications. First, it suggests there is a need for further study, including development and evaluation of intervention strategies promoting changes in social norms in the workplace around eating and PA. This work also suggests that employees may benefit from greater visibility for positive health behaviors, perhaps through management support. For example, managers might highlight a particular employee or group of employees’ participation in a positive fashion or encourage participation. Similarly, there may be benefit to rewarding/encouraging health behaviors so they become more pervasive as well as promotion of early adopters and champions. Finally, this work points to the potential for implementing and evaluating policy change at worksites aimed at creating a food and PA environment that fosters modeling. Such changes might include informal policies specifying healthy foods be served at workplace-sponsored events, such employees would observe each other eating healthy foods or that meetings should include exercise breaks, where employees might observe their peers being active. Future studies, using longitudinal and experimental designs should investigate these strategies.

This study has limitations worth noting. From this cross-sectional study, it is not possible to determine the direction of causality; for example if perceived organizational support for employee health causes lower BMIs among employees or if employees with lower BMIs might be drawn to such workplaces or may be those who are more aware of the supports an organization offers. Further, the relationship between norms and behaviors may suggest people notice those more like themselves, rather than suggesting that the diet and PA behavior of some co-workers influence those of others. Additionally, the workplace perceptions and measures for BMI, diet, and PA were collected by self-report, which are subject to bias as well as inaccuracy of reporting. Another limitation is the use of the IPAQ to assess physical activity. Specifically, it is not possible to assess the intensity of walking participants engaged in. For this analysis we included all walking as contributing to PA. Thus, we may have overestimated the PA level of the population; IPAQ overestimates PA when compared with objective measures [51, 52]. It is also worth noting that IPAQ was primarily developed for cross-national surveillance not estimating guideline adherence or intervention effects [41, 53, 54]. There is the potential for bias based on who responded to the survey, particularly given the low response rate and the use of only landline phone numbers. Further, the survey was conducted among working populations in Missouri, and therefore may not generalize to the non-working population.