Among US adults, only one-third eat enough fruits and a little over one-quarter eat enough vegetables per day to meet nutritional recommendations set in the Healthy People 2010 national objectives (1), and two-thirds are overweight or obese (2). Poor nutrition and obesity contribute to chronic diseases that result in billions of dollars in medical costs and lost work productivity per year (3-5). Forty-five percent of working-aged Americans have chronic diseases, including hypertension, diabetes, heart disease, stroke, and high cholesterol (6), that are affected by poor nutrition (3) and obesity (7,8). Some research suggests that fruit and vegetable intake is associated with having a lower body mass index (BMI) (9,10) and with reducing the risk for developing chronic disease (10-12). Research suggests that merely increasing the consumption of fruits and vegetables may help with weight management, and increased consumption of high-fiber, energy-poor fruits and vegetables often leads to a spontaneous reduction in fat intake (13). These assumptions about the health and weight management benefits of fruit and vegetable consumption need further investigation.

For decades, state-based programs have promoted public health messages about healthful eating. In recent years, state governments have recognized that their own state workers are as likely as the general public to have poor health habits that are associated with deleterious health outcomes and that directly influence state health care budgets (14). Studies on worksite wellness programs have shown that fruit and vegetable consumption increases when management and peers provide support, create environments that offer healthy choices, and reward participants with incentives (15-17). Consequently, many states have implemented worksite wellness initiatives to improve nutrition, promote physical activity, and reduce obesity among state workers (14,17).

Although states may develop wellness initiatives for state workers, few publish the findings regarding the implementation or evaluation of such programs. Among those few, the North Carolina HeartSmart study showed increased fruit and vegetable consumption among state employees after 1 year of program participation (17). The public workforce size makes program implementation difficult, but states such as North Carolina have initiated approaches (eg, using the Internet) for delivering wellness interventions to reach large numbers of employees (17). Computer or Web-based programs for nutrition promotion make it easier to provide education, monitor dietary intake, and track participant success in a cost-effective way (18). Such programs may provide feasible, affordable solutions to improve employee health in both public and private sectors and thereby reduce the effect of obesity and unhealthy dietary habits on employee health and health care costs.

The Healthy Employee Lifestyle Program (HELP) uses Web-based technology and site-specific program tailoring aimed at decreasing risk for chronic diseases and reducing health care costs among state employees in Arkansas. The Arkansas Department of Health (ADH) developed the HELP intervention and launched a 1-year pilot study in February of 2005. We report the analysis of the nutritional components of the 1-year pilot to assess the effectiveness of the intervention in promoting dietary changes among participants, including shifts in stages of readiness to change dietary practices.

Of the 10,000 ADH and DHS employees, 10% (n = 1,017) enrolled in HELP during the first year, February 2005 to March 2006. Most enrollees and participants were female and white (Table 1). Mean BMI of enrollees and participants was 30 kg/m², and approximately 75% of HELP enrollees and participants were either overweight or obese.

At follow-up, more participants ate 3 or more servings of vegetables per day than they did at baseline (26.2% vs 13.6%, P = .03) (Table 2). The data showed an overall trend of increased fruit consumption between baseline and follow-up. There was a shift in consumption of 3 or more fruits per day from 10.8% at baseline to 17.3% at follow-up (P = .08). Participants' consumption of the more healthful food groups of proteins, grains, and dairy did not increase significantly between baseline and follow-up. No significant differences toward decreased consumption of sweets/desserts, fats, fried foods, and processed meats were observed.

Participants progressed between baseline and follow-up in stages of readiness to change for eating 5 or more fruits and vegetables per day (P = .002) and for eating a low-fat diet (P = .04) (Table 3). For eating 5 or more fruits and vegetables per day, 42% of participants were in the preparation stage and 41% were in the action or maintenance stages. At follow-up, the percentage of participants in the preparation stage fell to 27% while the percentage in the action or maintenance stage increased to 59% (P = .002). Similarly, for eating a diet low in fat, 29% of participants were in the preparation stage and 49% were in the action or maintenance stage at baseline; at follow-up, the percentage of participants in the preparation stage fell to 21% and the percentage in the action or maintenance stage increased to 59% (P = .04).

Our findings from the 1-year pilot of the HELP program suggest that a Web-based worksite wellness incentive program may improve nutritional health behaviors of public-sector employees. The HELP program encourages behavior change through 3 main approaches: 1) providing an overall wellness report with tips for improving health, 2) rewarding health behaviors with points redeemable for incentives, and 3) providing education and peer support. These pilot results are promising, given the small sample of 214 participants. A larger sample would have been helpful in detecting more modest changes in behavior.

Fruit and vegetable consumption was the only dietary behavior rewarded by the HELP intervention. Other desirable behaviors (eg, decreasing consumption of fats or sweets) were not directly rewarded by the program. This reward system may have contributed to the lack of significant change in other dietary behaviors. Further investigation is necessary to determine how best to achieve change in the broader range of dietary behaviors.

Our findings are consistent with other wellness interventions that reported increased fruit and vegetable consumption and decreased fat consumption among program participants (16,17). For example, both the Seattle and Treatwell 5-A-Day worksite wellness studies reported increased fruit and vegetable consumption among participants (21,22). Similarly, the Worksite Internet Nutrition program reported decreased fat consumption and increased fruit and vegetable consumption among participants by using applied nutritional behavior change principles through an e-mailed intervention (23).

The changes observed among HELP participants may not be solely due to program participation. During the period of pilot implementation, Arkansas reformed state policies and organizational structures responsible for health programs in Arkansas, and media outlets heralded the state's multiple efforts to improve health by promoting nutrition and physical activity. Longer-term longitudinal studies are needed to determine the HELP program's effectiveness.

A limitation of this study was the inability to match precisely the baseline and follow-up HRAs for participants. Although the cross-sectional samples were matched closely, the interpretation of findings of change over time would have been strengthened by a true longitudinal sample. This study is limited further by the nature of the data, which were self-reported. The influence of this limitation may be mitigated somewhat by the failure of the program to reward change over time; therefore, there was little or no incentive for participants to skew their responses on the follow-up HRA in any single direction. Having 1 year between baseline and follow-up HRAs may have minimized the self-report bias because participants were not likely to remember previous responses. Further, to the extent that people in the matched samples were different from people for whom no match was possible, our ability to generalize these findings to program participants overall is limited. People who remained in the program and were motivated to complete a follow-up HRA could have been systematically different from those who failed to do so, further limiting the generalizability of these findings.

People with a high BMI tend to use health services more often, which lowers their work productivity because of absence and contributes to higher insurance premiums for employers (4,5,24,25). Most (75%) HELP enrollees reported BMIs in the overweight or obese ranges, suggesting that HELP was able to reach and recruit the desired, higher-risk employees. Our data do not include information on health care use or days missed from work, both of which would be indicators of program effectiveness.

The HELP pilot program produced positive outcomes in a brief period. Results suggest that the HELP intervention can be a feasible, affordable, easily implemented health behavior intervention that shows some promise for improving dietary behaviors of working adults. Findings suggest that, in time, risk, morbidity, and mortality may decrease if participants continue to increase healthy behaviors and decrease less healthy behaviors (5). Increased fruit and vegetable consumption is an easily communicated health message that shows promise for decreasing risk for chronic disease (17,21,22). Further analysis of the HELP data for physical activity and other health risk factors will be examined to determine whether the program merits a broader dissemination.