Who participates in internet-based worksite weight loss programs?
© You et al; licensee BioMed Central Ltd. 2011
Received: 26 May 2011
Accepted: 20 September 2011
Published: 20 September 2011
The reach and representativeness are seldom examined in worksite weight loss studies. This paper describes and illustrates a method for directly assessing the reach and representativeness of a internet-based worksite weight loss program.
A brief health survey (BHS) was administered, between January 2008 and November 2009, to employees at 19 worksites in Southwest Virginia. The BHS included demographic, behavioral, and health questions. All employees were blinded to the existence of a future weight loss program until the completion of the BHS.
The BHS has a participation rate of 66 percent and the subsequent weight loss program has a participation rate of 30 percent. Employees from higher income households, with higher education levels and health literacy proficiency were significantly more likely to participate in the program (p's < .01).
Worksite weight loss programs should include targeted marketing strategies to engage employees with lower income, education, and health literacy.
A recent systematic review on the effectiveness of worksite nutrition and physical activity programs for reducing weight in overweight and obese employees found a consistent, but modest, effect of interventions on reductions in body weight (i.e., -2.8 pounds) and body mass index (BMI; i.e., -0.5 BMI) when compared to untreated controls . These modest results could have a large public health impact if the intervention strategies could reach a large and representative or high risk sample of employees since the majority of adults spend a large amount of waking hours at work . Unfortunately, few worksite health promotion studies reported on reach,  defined as the proportion of the eligible employee population that participated in a program and the representativeness of program participants compared to the eligible population . Specifically, a review of worksite health prevention and intervention programs documented that only 25% of studies reported the proportion of eligible employees that agreed to participate in the study and only 9% reported on representativeness .
The current state of worksite intervention studies led the U.S. Preventive Services Task Force to conclude that there is a significant gap related to understanding the characteristics of the employee population that participates . Because of this gap, it is unclear if those who could benefit most from a worksite intervention are as likely to participate as those who may already be making more healthful lifestyle choices [3, 5]. This gap is also recognized in the general preventive care intervention research and several calls for better methods for reporting on external validity issues have been made [6–9] Understanding program reach not only would aid in filling the need of examining external validity of the program, but also would provide much needed information for determining whether or not the critical subgroups of employees are actually participating in the program. The results will also inform targeted marketing and recruitment efforts.
However, the documentation of reach and representativeness is predicated on knowing the proportion of eligible employees and having a clear picture of the demographic and behavioral characteristics of the total employee population, and this information has proven difficult to obtain. [e.g., [3, 10–12]] Therefore the purpose of this paper is to describe and illustrate a method for directly assessing the reach and representativeness of participants in a worksite weight loss study that focused on changes in physical activity and nutrition.
The data presented in this paper were collected as part of a two-group, cluster randomized controlled trial to investigate the reach and effectiveness of individually targeted, computer mediated worksite weight loss programs. The program comprises of emails with physical activity and diet messages that are tailored to the individual and includes small monetary incentives for weight-loss. Prior to program initiation a brief health survey (BHS) was administered to as many employees as possible at each participating worksite to determine eligibility, and evaluate reach and potential participation predictors. We administered the BHS between January 2008 and November 2009 with a goal to achieve a 70% completion rate prior to implementing the worksite weight loss intervention study. The analyses were completed in December 2009.
All employees with the exception of organizational decision-makers (i.e. CEOs and Human Resource Directors) were blinded to the existence of a future weight loss program to be delivered at their worksite, and as such employees' responses were unlikely to be biased by that knowledge. The BHS was introduced to the general employee population as a survey study to provide feedback on areas that the workplace health promotion efforts could focus on. The BHS was available in paper-and-pencil and web-based formats, and was delivered to all employees four weeks prior to the weight loss program being initiated. Employees had two weeks to complete the survey. All participating worksites allowed employees to complete the BHS during work hours. While it is noted in the literature that to achieve a 50 percent employee response rate to health risk surveys incentives of approximately $40 per employee are necessary when tied with strong organizational communication; providing such incentives was beyond the budget of the project. As a result, a lottery system was installed to encourage the survey participation. Any BHS participant could submit his/her name for a $250 prize drawing at each worksite. This study was approved by the Virginia Tech Institutional Review Board (protocol #07-296).
Worksites and Individual Employees Sample
Participation Rates By Worksite
% Eligible among those
completed the BHS
(BMI > = 25)
Rate among Eligible
Manufacturing and Distribution Centers
All employees were eligible to participate in the BHS. Employee eligibility criteria for the subsequent weight loss program include a BMI > = 25, regular employment status (e.g., temporary employees were excluded), and access to the internet. Therefore only those employees who were eligible to participate in the weight loss program are included in the analysis of this paper to determine weight loss program participant representativeness to the eligible population. The final analysis included 2,055 participants across 19 worksites ranging in size from 33-95 program participants and 100-589 total employees.
Brief Health Survey Data
As the primary outcome measure, we created a dummy variable indicating whether an employee enrolled in the subsequent weight loss program or not. Key information collected through the BHS includes: healthy eating, physical activity, health status, health literacy, internet and e-mail use proficiency, self-efficacy, response-efficacy, and demographic variables (i.e., age, gender, race/ethnicity, number of children in the household, household income, and education). We also controlled in our model other factors such as self-reported height and weight, smoking status, and existence of comorbid conditions. Height and weight were used to compute BMI. Smoking status was assessed using a single-item question: "Do you currently smoke?" Comorbid health conditions (arthritis, asthma, depression, diabetes, heart disease, high blood pressure, high cholesterol, obesity, and none of the above) were assessed using a single-item question which respondents checked all conditions that have been diagnosed by a doctor.
Reach and Representativeness Assessment
The initial reach metric calculated was the proportion of eligible employees (i.e., BMI > = 25) who enrolled in the weight loss program. The numerator for this analysis included all employees who enrolled in the study. The denominator used in this calculation was computed by multiplying the proportion of employees that responded to the BHS with a BMI greater than or equal to 25 by the total number of employees at the worksite to extrapolate to all eligible employees rather than just those who completed the BHS. Representativeness was calculated by comparing those eligible employees that participated in the program to eligible employees that did not participate in the program, but completed the BHS. In addition, there were a significant number of employees across worksites (n = 398) that participated in the program but did not complete the BHS. Thus we also compared these participants to those that completed the survey. This comparison allowed us to determine if those employees with a BMI ≥ 25 who responded to the BHS were representative of those who did not.
Weight Loss Program Reach Prediction
We used a multi-level mixed effect logit model treating the employees as nested within worksite, which is equivalent to estimating random intercept models assuming the unobserved worksite-specific effects are not correlated with predictors in the model. The model allows us to apply the study results to general worksite populations. The variety of worksites in the program, shown in Table 1, supports the population inference as well.
Reach and Representativeness Assessment
The BHS and the weight loss program reach rates are presented in Table 1. Across the 19 worksites enrolled in the study, the BHS participation rate on average is 66% with a median of 63% and ranged from 30% to 95%. For the weight loss program, the participation rate among the entire potentially eligible population ranged from 17% to 49% with a mean of 26% and a median of 23%.
Representativeness of Brief Health Survey (BHS) and the Weight Loss Program
Reach Study Sample
(n = 2,055)
BHS & Program Participants
(n = 610)
BHS Participants Program Nonparticipants
(n = 1,445)
BHS Nonparticipants Program Participants
(n = 398)
Group Mean Test p-value
Mean(2) = Mean(3)
Mean(2) = Mean(4)
Mean(1) = Mean(4)
Hispanic origin, %
Less than high school
High school graduate
Annual household Income, %
Less than $15,000
More than $100,000
Have at least one child, %
There were a total of 398 employees who did not complete the BHS but enrolled in the program. Those individuals were not in the program reach predication model analysis since their BHS data was missing. We compared their characteristics with the 610 program participants who completed BHS to assess the representativeness of our BHS sample. The seventh column of Table 2 presents the test results (i.e., comparing Mean(2) and Mean(4)). Majority of comparisons did not show statistically significant differences except for race/ethnicity which shows a relatively good representability of our BHS sample. However, those program participants who did not participate in the BHS on average were more likely to be Hispanic, or African American, and less likely to be Caucasian (p's < .01). To further assess the BHS representativeness, we compare those BHS participants with those 398 BHS nonparticipants. Results are shown in the last column of Table 2 (i.e., comparing Mean(1) and Mean(4)). It further confirms that our BHS sample is mostly representative except that female employees and those of Hispanic origin are less likely to fill out the BHS survey.
The results of the comparison test to assess the program participant representativeness are shown in the sixth column of Table 2 (i.e., comparing Mean(2) and Mean(3)). Statistically significant differences were detected comparing age, gender, proportion of Caucasian, African American, Asian, education level lower than some college, and income level lower than $30,000/yr.
Reach Prediction Model
Variable Descriptions and Summary Statistics across Sample and Worksites
N = 2,055
Dummy (= 1 if participate in the program; = 0 if not)
Body mass index score (kg/m2)
Rank (= 1 inactive; = 2 MSR or MCV; = 3 Meeting recommendations)
Healthy eating scores. Lower scores means healthier eating habits
Overall Health Status
Ranked self-reported health status (1 excellent to 5 poor)
Dummy (1 if has at least one comorbidity; = 0 if has no other health conditions other than obesity)
Dummy (= 1 if current smoker; = 0 otherwise)
Health literacy scores. Higher scores means better health literacy.
Self efficacy in internet use. Higher scores means higher efficacy.
Self efficacy in email use. Higher scores means higher efficacy.
Program Participation Multi-level Mixed Effect Logit Model Results
Overall Health Status
Comorbid Health Conditions
Race (Caucasian is the base)
Have at least one child
LR (mixed effect logit vs. normal logit)
Discussion and Conclusion
This paper describes an initial attempt at addressing the paucity in the research literature on the reach of worksite-based weight loss programs. Consistent with the literature, we found that older Caucasian women were more likely to participate in worksite weight loss programs [5, 12]. Of note, even though the weight loss programs were internet-based more women than men as well as older people were willing to participate suggesting that digital divide stereotypes may not apply in this context . Our study also adds to the literature and suggests that participants are more likely to have a chronic condition, higher BMI, and higher health literacy than the overweight and obese employees that choose not to participate.
A primary goal of this paper was to present a method to assess reach and representativeness in the context of worksite weight loss programs. While it would have been ideal to achieve a 100% completion rate of the BHS, the average completion rate of approximately two thirds of the employee population reflects a strong response rate relative to other worksite health risk appraisal completion data . This suggests that a lottery incentive system, when coupled with strong organizational communication strategies, could be as effective as individual level incentives. Further, our study also provides researchers a more cost effective method of achieving a greater than 50 percent BHS response rate.
We also found that African American and Asian employees were about one-half and one-quarter less likely than Caucasians to participate, respectively. Similar to the need to use culturally sensitive strategies within weight loss programs for diverse employee populations,[21, 22] our findings suggest the need to also develop culturally-tailored recruitment tactics and materials. For example, African Americans may be less likely to perceive themselves as overweight and more likely to associate attractiveness and health status with heavier body size when contrasted with Caucasians [23, 24]. For these reasons, programs that focus on outcomes aside from weight loss such as obesity-related co-morbid conditions or improved job performance and satisfaction may promote higher participation among African Americans. Similarly, employees with lower health literacy were found to be less likely to participate in the program which confirms the important implications of health literacy status in the context of recruiting and retaining participants [25, 26]. Our study directly assessed health literacy as a characteristic impacting enrollment which no known study has done before.
Limitations of this study include varying participation rates in surveys and an average BHS completion rate of approximately 66%. Although conducting a thorough examination of potential non-response bias was beyond the scope of this study, we did calculate adjusted reach estimates to account for these differences and provide comparison tests to assess BHS representativeness (Table 2). In contrast study strengths include a diverse set of worksites, a range of demographic, behavioral, and health literacy indicators; the study of both survey and program participation; and adjustments for clustering of employees within worksites.
The use of a BHS to provide information on employee characteristics and eligibility provides an opportunity to determine more accurate indicators of reach and representativeness. As reach and representativeness have received limited attention in worksite health promotion research, it is necessary to persist in collection and reporting of these types of data. The approach utilized in our study appears to be feasible and appropriate for use in future research.
We would like to thank all partners at each of the worksites that agreed to participate in the study. We would also like to acknowledge the work of Sarah Wall and Jennifer Parrish for coordinating and implementing the recruitment protocol necessary for this study. This study is supported by a grant from the National Institute for Diabetes and Digestive and Kidney Diseases: 5R01DK071664-04 (Estabrooks, PI).
- Anderson LM, Quinn TA, Glanz K, et al: The effectiveness of worksite nutrition and physical activity interventions for controlling employee overweight and obesity: a systematic review. Am J Prev Med. 2009, 37: 340-357. 10.1016/j.amepre.2009.07.003.View ArticlePubMedGoogle Scholar
- Estabrooks PA, Glasgow RE: Worksite interventions. Cambridge Handbook of Psychology, Health and Medicine. Edited by: Ayers S, Baum A, McManus C, Newman S, Wallston K, Weinman J, West R. 2007, New York: Cambridge University pressGoogle Scholar
- Bull SS, Gillette C, Glasgow RE, Estabrooks P: Work site health promotion research: To what extent can we generalize the results and what is needed to translate research to practice?. Health Educ Behav. 2003, 30: 537-549. 10.1177/1090198103254340.View ArticlePubMedGoogle Scholar
- Glasgow RE, Vogt TM, Boles SM: Evaluating the public health impact of health promotion interventions: the RE-AIM framework. Am J Public Health. 1999, 99: 1322-1327.View ArticleGoogle Scholar
- Linnan LA, Emmons KM, Klar N, Fava JL, LaForge RG, Abrams DB: Challenges to improving the impact of worksite cancer prevention programs: comparing reach, enrollment, and attrition using active versus passive recruitment strategies. Ann Behav Med. 2002, 24: 157-166. 10.1207/S15324796ABM2402_13.View ArticlePubMedGoogle Scholar
- Glasgow RE, Bull SS, Gillette C, Klesges LM, Dzewaltowski DA: Behavior change intervention research in healthcare settings: a review of recent reports with emphasis on external validity. Am J Prev Med. 2002, 23: 62-69. 10.1016/S0749-3797(02)00437-3.View ArticlePubMedGoogle Scholar
- Glasgow RE, Nelson CC, Strycker LA, King DK: Using RE-AIM metrics to evaluate diabetes self-management support interventions. Am J Prev Med. 2006, 30: 67-73. 10.1016/j.amepre.2005.08.037.View ArticlePubMedGoogle Scholar
- Glasgow RE, Emmons KM: How can we increase translation of research into practice? Types of evidence needed. Annu Rev Publ Health. 2007, 28: 413-433. 10.1146/annurev.publhealth.28.021406.144145.View ArticleGoogle Scholar
- Glasgow RE, Linnan L: Evaluation of theory-based interventions. Health Behavior and Health Education: Theory, Practice and Research. Edited by: Glanz K, Rimer B, Vishwanath V. 2008, San Francisco: Jossey-Bass Publishers, 4Google Scholar
- Franklin PD, Rosenbaum PF, Carey MP, Roizen MF: Using sequential email messages to promote health behaviors: evidence of feasibility and reach in a worksite sample. J Med Internet. 2006, Re. 8, e3Google Scholar
- Glasgow RE, Mccaul KD, Fisher KJ: Participation in worksite health promotion - a critique of the literature and recommendations for future practice. Health Edu Quar. 1993, 20: 391-408. 10.1177/109019819302000309.View ArticleGoogle Scholar
- Robroek SJW, Lenthe FJV, Empelen PV, Burdorf A: Determinants of participation in worksite health promotion programmes: a systematic review. Int J Behav Nutr Phy. 2009, 6: 26-10.1186/1479-5868-6-26.View ArticleGoogle Scholar
- Taitel MS, Haufle V, Heck D, Loeppke R, Fetterolf D: Incentives and other factors associated with employee participation in health risk assessments. J Occup Environ Med. 2008, 50: 863-72. 10.1097/JOM.0b013e3181845fe2.View ArticlePubMedGoogle Scholar
- Ammerman A, Haines P, DeVellis R, et al: A brief dietary assessment to guide cholesterol reduction in low income individuals: design and validation. JADA. 1991, 91: 1385-1390.Google Scholar
- Macera CA, Pratt C: Public health surveillance of physical activity. Res Q Exercise and Sport. 2000, 71: 97-103.View ArticleGoogle Scholar
- Scientific Advisory Committee of the Medical Outcomes Trust: Assessing health status and quality-of-life instruments: attributes and review criteria. Quality of Life Research. 2002, 11: 193-205. 10.1023/A:1015291021312.View ArticleGoogle Scholar
- Chew LD, Bradley KA, Boyko EJ: Brief questions to identify patients with inadequate health literacy. Fam Med. 2004, 36: 588-594.PubMedGoogle Scholar
- Schwarzer R, Renner B: Social-cognitive predictors of health behavior: action self-efficacy and coping self-efficacy. Health Psychology. 2000, 19: 487-495.View ArticlePubMedGoogle Scholar
- Plotnikoff RC, Higginbotham N: Protection motivation theory and exercise behaviour change for the prevention of coronary heart disease in a high risk, Australian representative community sample of adults. Psychology, Health and Medicine. 2002, 7: 87-98. 10.1080/13548500120101586.View ArticleGoogle Scholar
- Wilson JJ, Mick R, Wei SJ, et al: Clinical trial resources on the Internet must be designed to reach underrepresented minorities. Cancer J. 2006, 12: 475-481. 10.1097/00130404-200611000-00007.View ArticlePubMedGoogle Scholar
- Kumanyika S, Whitt-Glover M, Gary TL, et al: Expanding the obesity research paradigm to reach African American communities. Prev Chronic Dis. 2007, 4: A112-PubMedPubMed CentralGoogle Scholar
- Zapka J, Lemon SC, Estabrook B, Rosal MC: Factors related to weight loss behavior in a multiracial/ethnic workforce. Ethn Dis. 2009, 19: 154-160.PubMedPubMed CentralGoogle Scholar
- Johnson-Taylor WL, Fisher RA, Hubbard VS, Starke-Reed P, Eggers PS: The change in weight perception of weight status among the overweight: comparison of NHANES III (1988-1994) and 1999-2004 NHANES. Int J Behav Nutr Phys Act. 2008, 5: 9-View ArticlePubMedPubMed CentralGoogle Scholar
- Thomas A, Moseley G, Stallings R, Nichols-English G, Wagner PJ: Perceptions of obesity: black and white differences. J Cult Divers. 2008, 15: 174-180.PubMedGoogle Scholar
- Frew PM, Rio C, Lu L, Clifton S, Mulligan MJ: Understanding differences in enrollment outcomes among high-risk populations recruited to a phase IIb HIV vaccine trial. J Acquired Immune Deficiency Syndromes. 2009, 50: 314-319. 10.1097/QAI.0b013e3181945eec.View ArticleGoogle Scholar
- Mancuso CA, Rincon M: Impact of health literacy on longitudinal asthma outcomes. J General Internal Med. 2006, 21: 813-817. 10.1111/j.1525-1497.2006.00528.x.View ArticleGoogle Scholar
- Lee S, Brown ER, Grant D, Belin TR, Brick JM: Exploring nonresponse bias in a health survey using neighborhood characteristics. Am J Public Health. 2009, 99: 1811-1817. 10.2105/AJPH.2008.154161.View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2458/11/709/prepub