The high prevalence of obesity is a major public health problem because of the association of obesity with chronic health conditions such as coronary heart disease, type 2 diabetes, and some cancers (1). The increase in prevalence during the last several decades more likely stems from environmental factors at multiple levels than genetic changes (2). More than one-third of the American adult population is obese (3); long-term behavior-change interventions that focus on weight reduction have had limited success in reducing overweight and obesity. From a socioecological perspective, behavioral interventions may have limited success because they inadequately consider environmental factors (4).

Research interest in the potential role of stress in health is growing. Stress is an individual-level factor associated with environmental phenomena and individual-level disease processes (5). Stress occurs when environmental demands tax or exceed the adaptive capacity of an organism; the demands result in physiologic or psychological processes that put the organism at risk for disease (5). Adults experience numerous types of stress (eg, work, finances, family), and each type contributes to overall stress. Although types of stress may vary from person to person and group to group, overall measures of stress allow generalized comparisons. Measures of stress are imperfect, but measures of perceived stress are valuable because they account for differences in the appraisal of what is stressful, exposure to stressors, and coping ability (6).

Perceived stress is associated with direct changes to both physiologic (eg, hormonal response) and psychological processes. Chronically elevated levels of perceived stress affect cortisol levels, which have been associated with increased risk for central obesity (7). Evidence for an association between stress and physical activity behaviors is mixed (8), and more testing of physical activity theory is needed to identify inconsistencies in the literature (4,9). Experimental studies have demonstrated that acute stress affects dietary behaviors, especially among people with certain eating behaviors, such as restrained eating (ie, intentional caloric restriction) (7,10). The way in which food is consumed has also been associated with overweight and obesity in epidemiologic settings (11,12). Such behaviors include emotional eating (ie, eating to manage feelings) and task eating (ie, eating while doing other activities). These behaviors connote a level of distraction — a lack of eating awareness — that has also been linked to obesity (13). A lack of eating awareness may co-occur with a hectic or stressful lifestyle and facilitate unhealthy dietary choices.

Some types of stress, such as work stress, have been associated with obesity-related behaviors among adults (14-16). Evaluating the relationship between perceived stress and obesity-related behaviors in a working population may be especially valuable because this population likely experiences numerous types of stress and a broader range of stress exposure (17,18). More than half of adults in the United States and in King County, Washington, our study population, are employed; they typically spend more than half of their waking hours at work (19). The objective of this study was to examine associations between perceived stress, dietary behavior, physical activity, self-efficacy, and body mass index (BMI) among healthy working adults. Secondary objectives were to explore whether eating awareness modified the relationship between perceived stress and dietary behavior. We hypothesized that behavioral and self-efficacy indices of dietary behavior and physical activity would be associated with overall perceived stress (Figure).

Among PACE participants, 82.7% were non-Hispanic white; almost half were college graduates and had a household income of $75,000 or more (Table 1). The mean (SD) PSS-10 score among all participants was 12.7 (6.4), and the mean (SD) BMI was 29.2 (6.3). A higher level of stress was associated with a higher percentage of participants who engaged in task eating always or most of the time, a lower Godin leisure-time exercise score, a lower percentage of participants who regularly engaged in sweat-inducing exercise, and fewer minutes walked per week (Table 2). We found a significant interaction between perceived stress and eating awareness in relation to the 5 A Day (z = −3.01, P = .003), single-item summary of fruit and vegetable consumption (z = −2.77, P = .006), and fast food consumption (z = 3.00, P = .003). Among participants who had a low level of eating awareness, higher levels of perceived stress were associated with fewer servings of fruit and vegetables per day and greater consumption of fast food meals (Table 3). We also found a significant interaction between perceived stress and eating awareness in relation to soft drink consumption (z = 2.97, P = .003); associations between perceived stress and soft drink intake in stratified analyses, however, were not significant.

Our sample of working adults was similar to US adults in terms of BMI (1) and perceived stress (30). Compared with the US population, however, our sample had a higher level of income and education and a higher proportion of non-Hispanic whites (11). BMI in our sample was not associated with perceived stress. The distribution of perceived stress scores, however, indicated average levels of stress (ie, levels experienced in everyday living) and does not reflect a highly stressed population (eg, shift-workers, racial/ethnic minority groups) where we would expect to find an association (31). In other words, the range of perceived stress scores in our sample setting was likely too low to detect an association. However, this same range was sufficient to detect associations with more proximal obesity-related behavioral indices.

Higher frequency of task eating was associated with higher levels of perceived stress, but overall dietary behavior and perceived stress was not associated. We also did not detect a sex interaction in the relationship between perceived stress and dietary behavior and physical activity. The relationship between perceived stress and dietary behavior may be modified by several factors, including overweight or obesity, sex, and domains of eating behavior (eg, restrained or emotional eating) (32). Women and overweight or obese people may be more likely to eat in ways in which dietary choice is influenced by stress. In controlled laboratory studies, participants, particularly women and restrained eaters, have responded to high levels of stress by consuming foods high in calories, sugar, and fat (ie, more palatable foods), fewer main meals, and fewer portions of vegetables (7,10,33,34).

Growing evidence suggests that stress’s influence on what you eat depends on how you eat. People characterized as stress- or emotional-eaters tend to choose calorically dense foods to blunt their stress response or reduce negative emotions (7). The relationship between restrained eating and stress is more complex; retrained eaters tend to eat less during normal conditions and overeat when stressed (7). We propose that variables used to operationalize these eating behaviors share the underlying construct of lack of eating awareness. That is, when energy-dense eating serves as reward during times of stress or negative emotions, assessment and moderation (ie, awareness) of caloric intake is absent. The interaction of low eating awareness on the association between stress and dietary behaviors in our data supports this line of thinking.

Higher levels of perceived stress were also associated with lower levels of physical activity. These findings are consistent with those found in most previous studies (8), although the direction of the relationship has not been determined. Stress may inhibit people from engaging in physical activity, or lack of physical activity may lead to increased stress levels, or both of these scenarios may occur. Numerous cross-sectional studies document both a presence and lack of an association between higher levels of stress and lower levels of physical activity (8). Of the null studies, however, two (35,36) measured perceived stress among older persons, who tend to have lower levels of stress (30), which decreased the likelihood of detecting an association. A pos thoc analysis (37) found that stress was more strongly associated with physical activity among people aged 30 to 44 (a group hypothesized as having higher levels of stress) than among people aged 18 to 29 or 45 or older. Higher levels of stress also predicted lower levels of physical activity among a prospective cohort of people at risk for diabetes (38).

Higher levels of stress have been associated with a lack of adherence to physical activity (35). Among a study of postmenopausal women, stress was modestly associated with failure to return to activity among those who had stopped (35). Studies on middle-aged women (39) and university students (39) found a decrease in exercise duration and an increase in the number of missed exercise opportunities as stress increased. Self-efficacy for dietary behavior and physical activity has also been negatively associated (but not significantly) with perceived stress (8,38). These findings support the idea that stress inhibits positive health behaviors, ultimately affecting one’s risk for obesity; they are also consistent with studies on stress and obesogenic behaviors among adults. Lower levels of social support among workers were associated with lower fruit and vegetable intake and lower levels of leisure-time physical activity (14), whereas greater work demands were associated with lower levels of physical activity (16), higher levels of task eating (women only) (15), and higher levels of fast food consumption (men only) (15) among working adults.

This study has several limitations, including its secondary cross-sectional design and the potential issue of multiple comparisons. Confirmatory analyses are needed, especially within a longitudinal setting. We did not include other measures of stress-related eating behaviors (eg, emotional eating) because evaluating the relationship between stress and eating was not the primary objective of the PACE intervention. Alcohol consumption, a potentially confounding variable in the relationship between stress and BMI, was also not measured (22). Using single items may also have been a disadvantage and perhaps relevant to our measures of self-efficacy. It may be important, for example, to measure both confidence in and barriers to self-efficacy because each may relate to different aspects of adopting behaviors in this population (4,9). Finally, most of our data were based on self-report. By using behavioral indices, we sought to minimize the biases potentially introduced by self-report and decrease the burden on the participant. Strengths of the study included adjustment for the multilevel data, analyses accounting for variance at worksite and individual levels, and use of physical measurements for BMI calculation.

In this study, perceived stress was associated with several obesogenic behaviors among mostly non-Hispanic white working adults with average levels of everyday stress. Our data provide insight into the potential role of even average levels of everyday stress in dietary, eating, and physical activity behaviors. The association of perceived stress and these behaviors is likely greater in people who have higher levels of stress. Further exploration of the role of stress in occupational groups in which excessive obesity-related behaviors have been documented may benefit future intervention strategies. Our findings may also suggest that including stress management and/or mindfulness techniques in worksite behavior-change interventions could improve program effectiveness.