The DASH for Health program was offered as a free employee benefit to all US-based employees (approximately 12,500) of EMC Corporation, a global information infrastructure company based in Hopkinton, Massachusetts. The program was also offered to all adult household members of these employees. Employees and household members were encouraged to join the online program through a series of email communications from EMC leaders. During a three-week open enrollment period, 3479 subjects enrolled in the program and logged on to the website at least once. At the time of enrollment, we asked enrollees if we could use information that they entered about themselves on the website (eg, weight, blood pressure (BP) levels, food intake) to determine whether the program was providing benefit. This report is based upon the 2834 enrollees (81%) who granted consent.

The project was approved by the Institutional Review Board of Boston University Medical Center.

Enrollees were given access to a personalized, password-protected website. Figure 1 displays a view of the program’s homepage as designed for the EMC audience. Through that website, we published new articles once a week. The articles contained information about elements of healthy nutrition. About 15% of the articles also dealt with healthy exercise practices. A sub-article each week addressed specific issues to promote weight loss. Articles were typically published every Friday and a reminder email was sent to each enrollee at the time a new article was posted on the website. The emails contained a brief description of that week’s article and a hyperlink to the log-in website. Nutrition advice was based on the DASH Diet, meaning that we instructed people about consuming various servings of the eight DASH food groups (fruits, vegetables, low-fat dairy, meat/fish/poultry, grains, nuts/legumes, sweets, and added fats). The articles were not targeted to specific subsets of the enrollee population. All enrollees received the same articles.

Figure 1

A view of the program’s homepage as designed for EMC Corporation employees

We had three primary outcomes, all measured at 12 months: first, change in weight between baseline and 12 months in subjects who indicated a desire to lose weight on their enrollment questionnaire; second, change in systolic blood pressure (SBP) in those who indicated that they either had high blood pressure or were on blood pressure medications or had been told to watch blood pressure (we defined this as our “High Blood Pressure” group); and third, change in consumption of DASH food groups. Change in diastolic blood pressure (DBP) was a secondary outcome. We also performed exploratory analyses of the relationship between our outcomes and the amount of use of the DASH for Health website.

For weight measurement, we used self-entered weights from the website which may have included weights taken by the subjects themselves or taken in other settings (eg, physician visits). We classified subjects as overweight/obese based on their body mass index (BMI; kg/m²). Similarly for blood pressure, we used self-entered readings which could have been self-measurements or readings taken by others. We did provide recommendations on the website about how to take one’s own blood pressure (seated, left arm, average of two measurements). In addition, the employer, EMC Corp., offered free automated sphygmomanometers (Fore-Care 6400; Forecare Inc., Buffalo Grove, IL) to enrollees with hypertension. Food consumption was estimated from the DASH Online Questionnaire, a 24-hour recall instrument. For weight, blood pressure, and food intake, if there were more than a single entry during the baseline or 12th-month time window, we averaged the entries and used that single value in our analyses. Website use was calculated as the number of log-ons by each enrollee who visited the website (unique users).

For our analyses, we used the data that enrollees self-entered on the website. There was no randomization and no control group. Our analyses do not allow estimation of what the effects of DASH for Health might have been on enrollees who did not enter any data. The baseline data reflect information that enrollees entered on the DASH for Health website during the three-week enrollment period (before the website was delivering any behavior-changing messages), and the “12th-month” data are those entered during weeks 48-52. The number of subjects analyzed for each outcome was determined by the number of subjects who entered data for that outcome during both the baseline and 12th-month time frame. In analyzing website use, we used the number of log-ins over 12 months. Data are displayed as means unless otherwise noted, and indices of dispersion are standard deviation (SD) or 95% confidence intervals (CI) as noted. All analyses were performed with SigmaStat 3.5. Baseline versus 12th-month comparisons were performed with paired t-tests or Wilcoxon signed rank tests. Statistically significant results had P values less than .05.

Enrollees were widely distributed geographically, residing in 41 states. They were approximately evenly distributed by gender, and their ages ranged from 18-73 years (average 40.7 years). They were highly educated, with 1845 (66%) having completed college or postgraduate work. Of the subjects, 88% were white, and 74% were married (see Table 1 for absolute numbers of subjects). The most commonly stated reasons for enrolling in DASH for Health were desires for general health information and weight loss. Approximately 25% were also concerned about blood pressure or cholesterol levels. Other demographic details are shown in Table 1.

Of the 3479 subjects who enrolled in the program and logged on to the website at least once, 2834 (81%) granted consent to use their data for research purposes. Of these, 735 (26%) were still actively using the website in the 12th month. Their demographics are also shown in Table 1. The groups were comparable except that subjects in the 12-month group were older, included a greater percentage of women, had fewer single and more widowed subjects, and had a greater percentage who were interested in “general health information” at the time of enrollment.

At the end of 12 months, 735 of the original 2834 enrollees (26%) were still actively visiting the website. Figure 2 displays the pattern of website use during a 3-week baseline period and then in sequential 4-week periods for 12 months. The dropout rate was highest during the first 2 months. The number of users stabilized and remained fairly constant for the final 5 months. On average, users visited the website two to three times during any 4-week period. However, it was not the same group of users who visited the website during each 4-week period: enrollees were dropping out and dropping back in over the entire 12 months.

Figure 2

Pattern of website use during a 3-week baseline period and then in sequential 4-week periods for 12 months

There were 203 subjects who indicated a desire to lose weight when they enrolled in the program and who entered their weight during the baseline period and during the 12th month of the program. Their average weight change was -3.1 lbs (CI -4.7, -1.5; P < .001). The overall range of weight change was +30 to -62 lbs. Of these 203 subjects, 151 had BMI in the overweight/obese range, ranging from 25.0 to 50.6 kg/m². This was the subgroup for our weight-loss primary outcome. Their average weight change was -4.2 lbs (CI -6.2, -2.2; P < .001). Of these, the 74 obese subjects (BMI ≥ 30) lost 5.2 pounds while the 77 overweight subjects (BMI ≥ 25 to 29.9) lost 3.4 pounds (weight change in obese versus overweight, P = .63). For the remaining 53 subjects, with BMI 18.5 - 24.9 kg/m², the average weight change was +0.2 lbs. Other characteristics of these subjects are shown in Table 2.

A total of 120 subjects entered blood pressure readings on the website during the baseline period and the 12th month (Table 3 and Figure 3). Of these, 62 met the definition of our high blood pressure group. Their systolic pressure change was -6.8 mmHg (P < .001); diastolic change was -2.1 mmHg (P = .16). An additional 58 subjects who indicated no blood pressure concern on their baseline questionnaires also entered blood pressure recordings in the baseline and 12th month. Their baseline blood pressure was lower than in the high blood pressure group (Table 3), and their systolic and diastolic pressure change was -2.4/-0.2 mmHg (P = .09 and .90, respectively).

Figure 3

Mean (+ /- 95% CI) systolic and diastolic blood pressure change at 12 months in high and normal blood pressure groups (the systolic change in the high blood pressure group was significant, P < .001)

A total of 181 enrollees completed at least one DASH online food questionnaire during the baseline period and the 12th month. The median number of completed questionnaires per enrollee was three during baseline and three during the 12th month. The average age was 42.4 years; 107 were women; 74 were men. Table 4 displays the average DASH goal (as servings / day) for each of the eight DASH food groups, as well as the number of servings consumed (as entered in the DASH online questionnaire). Consistent with DASH recommendations, there were significant increases in daily fruit and vegetable intake in the group. There was also a significant decrease in the consumption of grain products, moving counter to the DASH goal. There were no significant changes in any of the other five DASH food groups.

The DASH online questionnaire also provided information on 52 subcategories of these eight main food groups. We performed exploratory analyses to examine changes in these subcategories. There were several significant changes in subgroup consumption. Three categories merit mention. Consumption of carbonated beverages decreased from 9 oz per day to 6.5 oz (P < .001). Enrollees reported reducing their consumption of refined-wheat bread products by 0.4 servings per day while increasing consumption of whole grain bread products by 0.3 servings per day (P < .001 and P = .004, respectively). So, while grain consumption overall decreased, the shift from refined grains to whole grains is consistent with DASH advice.

We performed exploratory analyses, relating the amount of website use (measured as number of log-ins over the course of 12 months) versus change in our main outcomes: weight, systolic blood pressure, and consumption of DASH food groups. We divided the sample into two parts based on the median number of log-ins. The median log-in number differed for each outcome, being determined by the number who provided baseline and 12-month data for that outcome. For weight and blood pressure, there were tendencies toward greater effect among those with more log-ins (Table 5). The only significant difference in food group consumption (data not shown) was greater fruit intake among those with a greater number of log-ins (P = .03). There were no differences in other food groups. The median log-in number for the group who provided dietary data (food questionnaires) was 44.