We worked with a large health care system and 2 local chapters of a national union representing retail workers to recruit participants. The health care system and the union represent large, fast-growing segments of the low-wage workforce, and both expressed interest in improving their workplace wellness efforts. Workers were recruited and surveyed from November 2013 through June 2014. We targeted hospital departments with high proportions of low-wage workers, including housekeepers, food service workers, patient care technicians, and unit secretaries; retail workers were primarily employed by 3 regional retail chains. We attempted to recruit all workers within targeted departments, stores, or union meetings and worked with supervisors, store managers, and union leaders to distribute paper surveys packets. Packets included a recruitment letter, consent form, and survey. Participants could return paper surveys in person to a research team member at a specified time and location or by mail using a prepaid envelope; they were compensated for their time. A small number of surveys were offered online to hospital employees who did computer work. All participants were at least 18 years of age and spoke English. This study was approved by the Washington University Institutional Review Board.

The survey assessed various domains including demographics, job characteristics, and work environment (eg, schedule, wages, social support, employer’s value of workers’ health), availability of and participation in WHPs, health behaviors (eg, diet, physical activity, willingness to change health behaviors), and health status (eg, height, weight, overall health, health conditions). To measure the 3 primary outcomes of diet, physical activity, and participation in WHPs, we used well-established survey tools: the Rapid Eating Assessment for Participants Short Version (REAP-S) (17), the 2-question physical activity assessment (18), and 16 items from the Worksite and Energy Balance Survey (WEBS) to measure availability and participation in WHPs (such as health fairs, exercise programs and facilities, flexible time for physical activity, and incentives to walk or bike to work) (19). The survey also included the SF-8 to measure health status (20), the Supervisor and Coworker Support scales from the Job Content Questionnaire (21), questions from the WEBS survey to determine food source at work, a question from the National Health and Nutrition Examination Survey Occupation Questionnaire Section (22) to assess work schedule, and a revised version of the Stanford Brief Activity Survey to determine physical activity at work (23).Willingness to change eating behaviors was measured by using a question from the REAP-S; we created a similar question for physical activity. Prior to distribution, we conducted pilot testing of the survey among both hospital and retail workers to ensure clarity, relevance, and readability. The survey took approximately 15 to 20 minutes to complete.

Body mass index (BMI) (weight in kg/height in m²) was calculated by using self-reported height and weight. Aggregate scores for the Job Content Questionnaire and the SF-8 physical and mental well-being scores were calculated by using published procedures (20,21). To assess food sources at work, participants reported the number of days that they brought food from home, purchased food at their workplace, or purchased takeout food to eat at work. Since workers often brought and purchased food on the same day, we assessed food sources in 2 ways: 1) we categorized the primary food source as the source of food more than 60% of the time, and 2) we calculated the proportion of time workers used each source. Work groups were compared using Pearson χ² and analysis of variance (ANOVA); significance was assessed at P ≤.05.

We examined possible predictors of 3 outcomes: diet, exercise, and WHP participation. To assess diet, we used the REAP-S total score (17), which reflects how often a participant engages in healthy and unhealthy eating behaviors. Scores ranged from 13 to 39 with a lower score indicating healthier behaviors. We also examined REAP-S subscores for consumption of fatty foods and sugary foods. To categorize people as either meeting or not meeting the recommended level of exercise (24), we estimated total physical activity minutes per week on the basis of answers to the 2 physical activity questions. Participation and availability of WHPs was calculated as a positive response for participation in any 1 of the 16 programs queried.

Student’s t test was conducted on dichotomous predictors of REAPS-S total score, and Spearman correlations were conducted for interval and ordinal predictors. Univariate logistic regression was conducted for predictors of exercise and WHP participation, yielding odds ratios and 95% confidence intervals. For each outcome, significant predictors (P ≤.05) in the univariate analyses were included in multivariate models. We analyzed the REAP-S total score by using multivariate ordinary least squares regression and used multivariate logistic regression to analyze exercise and WHP participation. Analyses were conducted using IBM SPSS version 20 and R version 3.1.0 (IBM Corporation).

The overall population had a REAP-S total score of 25.5 (SD, 4.5), indicating that many respondents had unhealthy eating habits. Compared with nonobese participants, obese participants had significantly higher scores on the REAP-S (25.1 vs 26.0, P = .04) and the fatty foods subscale (8.0 vs 8.4, P = .047); there was no significant difference in the sugar subscale. The source of food at work varied greatly; about 23.3% primarily brought from home, 38.8% primarily bought food at work, 24% split between bringing and buying, and 13.1% did not regularly eat at work.

Overall, only 30.9% reported getting the recommended level of exercise, lower than the 46.1% found in a national sample (26). Obese workers were less likely to get recommended levels of exercise than nonobese workers (23.8% vs 35.8%, P = .006). More than one-third (35.7%) reported spending most of their work day either sitting or standing, whereas 28.7% said they spend most of the day walking, 28.9% said they spend most of the day lifting or pushing heavy objects or moving most of their body, and 4.3% said they do hard physical labor most of the day.

Most participants reported willingness to change both eating habits and physical activity to be healthier (reporting at least a 4 on a scale of 1 to 5 with 1 being “not at all willing” and 5 being “very willing”); many said they had already changed eating patterns or physical activity in the last year because of health concerns.

More than half (55.8%) of respondents did not regularly work day shifts, and 32.4% reported working irregular schedules. Overall, participants felt that their supervisors and coworkers were supportive as indicated by high Supervisor and Coworker Support scales scores. Additionally, most workers agreed or strongly agreed that their companies valued healthy workers. Participation in any WHP was 36.7%; among those who reported that WHPs were offered, the participation rate was 54.8%. Availability of WHPs was not associated with lower rates of obesity, but those who participated in 1 or more programs were less likely to be obese than those who did not (49.7% vs 60.7%).

In univariate analyses, a lower REAP-S score, (ie, healthier diet) was associated with older age, higher wages, greater number of hours worked, higher rate of bringing food from home, having some college education, participating in a WHP, and working for the hospital system rather than for retail stores (Table 2). Minority status, nonday shifts, irregular shifts, and higher rates of buying food at work or getting takeout were associated with unhealthier diet. The final multivariate models had a R² value for the REAP-S total score of 0.24 for all workers (Table 3). Bringing food from home was the strongest predictor of healthy diet for all workers. Older age, lower wages, nonminority status, some college education, and participation in a WHP were also predictors of healthy diet.

Fewer predictors of exercise were found via univariate analysis (Table 2). For all workers, younger age, higher rate of bringing food, lower rate of buying food, having more physical activity at work, and participating in WHPs were all significant predictors of exercise. In the multivariate logistic regression model, only physical activity at work and WHP participation were significant predictors of exercise for all workers (Table 3).

Univariate logistic regression analyses of the 354 workers who indicated that their company offered 1 or more WHPs showed that younger age, being female, being a minority, and working for the hospital predicted WHP participation, whereas working nonday shifts and having irregular schedules were associated with nonparticipation (Table 2). In the multivariate model for all workers, younger age, minority status, and being a hospital worker predicted participation in WHP.

Several differences between the work groups may inform future interventions. Table 1 shows comparisons between hospital workers and retail workers. Retail workers were more likely to work nonday shifts, have irregular schedules, and sit or stand in 1 place most of the day compared with hospital workers (Table 1). Hospital workers were more likely to believe that their company valued healthy workers than retail workers and also reported greater availability of WHPs. Significant (P ≤.05) univariate associations with diet, exercise, and WHP participation for hospital workers and retail workers are noted in Table 2.

The R² values for the multivariate models predicting the overall REAP-S score were 0.26 for hospital workers, and 0.22 for retail workers. For hospital workers, bringing food from home and nonminority status were associated with a healthier diet; bringing food from home, participation in a WHP, younger age, and nonminority status were predictors of healthier diet in retail workers (Table 3).

In the multivariate models, WHP participation was the only significant predictor of exercise in retail workers (odds ratio [OR] 2.16, P = .03); there were no significant predictors of exercise for hospital workers. Although participation in any WHP was not a significant predictor of exercise in hospital workers, participation in 4 programs was associated with exercise in these workers: workplace exercise programs (OR, 3.12; P = .04), reduced price gym memberships (OR, 4.30; P = .008), signs encouraging the use of stairs (OR, 4.35; P = .02), and brochures or a poster encouraging healthy behaviors (OR, 3.00; P = .009).