The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
A Randomized Controlled Trial of a Commercial Internet Weight Loss Program
Article first published online: 6 SEP 2012
2004 North American Association for the Study of Obesity (NAASO)
Volume 12, Issue 6, pages 1011–1018, June 2004
How to Cite
Womble, L. G., Wadden, T. A., McGuckin, B. G., Sargent, S. L., Rothman, R. A. and Krauthamer-Ewing, E. S. (2004), A Randomized Controlled Trial of a Commercial Internet Weight Loss Program. Obesity Research, 12: 1011–1018. doi: 10.1038/oby.2004.124
- Issue published online: 6 SEP 2012
- Article first published online: 6 SEP 2012
- Received for review November 06, 2003; Accepted in final form April 26, 2004
- weight management;
- lifestyle modification;
- behavior therapy;
- e-mail counseling;
- Internet diets
- Top of page
- Research Methods and Procedures
Objective: To assess, in a 1-year randomized controlled trial, the efficacy of eDiets.com (a commercial Internet weight loss program) in improving weight, cardiovascular health, and quality of life.
Research Methods and Procedures: Participants were 47 women with a mean age of 43.7 ± 10.2 (SD) years and a mean BMI of 33.5 ± 3.1 kg/m2. They were randomly assigned to either: 1) eDiets.com, a commercial Internet-based program available to the public; or 2) a weight loss manual (i.e., LEARN Program for Weight Control 2000). At baseline, participants in both groups met briefly with a psychologist who instructed them to follow the components of their program as closely as possible. Additional brief visits were provided at weeks 8, 16, 26, and 52 to review their progress. Change in weight was the main outcome measure.
Results: At week 16, participants in eDiets.com lost 0.9 ± 3.2% of initial weight compared with 3.6 ± 4.0% for women assigned to the weight loss manual. At week 52, losses increased to 1.1 ± 4.0% and 4.0 ± 5.1%, respectively. Results of a last-observation-carried-forward analysis found that women in the manual group lost significantly (p < 0.05) more weight (at both times) than those treated by eDiets.com. (Results, however, of baseline-carried-forward and completers analyses did not reach statistical significance.) There were no significant differences between groups in changes in cardiovascular risk factors or quality of life.
Discussion: This study provides consumers with important information about the probable benefits they can expect from participating in a popular Internet-based weight loss program.
- Top of page
- Research Methods and Procedures
Nearly two-thirds of the U.S. population is overweight or obese (1), making the discovery of successful weight loss interventions an important public health issue. Group behavioral weight loss programs are among the most widely researched treatments and consistently induce an 8% to 10% reduction in initial weight in 6 months (2,3,4). These programs, however, are time intensive, requiring weekly on-site meetings. Time and travel demands may contribute to the typical attrition rate of 20% in 4–6 months, with greater attrition in longer programs (2,3).
Recent research has investigated interventions designed to decrease the demands of on-site behavioral programs. Such efforts have included the use of telephone and interactive television (5,6,7). Several studies also have shown that behavioral weight control can be successfully provided through Internet and e-mail contact (8,9,10). In addition to being accessible 24 hours a day, the Internet offers anonymity that could possibly encourage obese individuals who are embarrassed about their weight to seek treatment. Interventions with reduced time and travel demands also could make treatment more available to the millions of Americans who need it.
Several commercial Internet programs are now offered to the public (11), including Weight Watchers (12), Jenny Craig (13), Nutrisystem (14), Diet Smart (15), andeDiets.com (16). None of these programs has reported results of its Internet-based program as determined in randomized controlled trials. This study assessed weight losses over 1 year for overweight and obese individuals who were randomly assigned to eDiets.com, a popular commercial Internet weight loss program. Weight losses of these participants were compared with those of individuals who received a widely used weight loss manual. Because participants in the eDiets.com group could receive unlimited social support from on-line meetings and bulletin board support groups, we predicted that they would lose significantly more weight than those who received the manual. This study represents the first investigation of a commercial Internet weight loss program. Findings from such studies are critical to the public's making informed choices about commercial weight loss programs, as called for by the Federal Trade Commission (17).
Research Methods and Procedures
- Top of page
- Research Methods and Procedures
Forty-seven women with a mean age of 43.7 ± 10.2 (SD) years and a mean BMI of 33.5 ± 3.1 kg/m2 were recruited by telephone calls from research assistants who described the purpose of the study and the two treatment groups (see Figure 1). Potential participants were identified from a pool of respondents to public service announcements and newspaper advertisements that described a variety of weight loss programs offered at the University of Pennsylvania. Participation was limited to women, ages 18–65 years, who had a BMI of 27–40 kg/m2. Individuals had to be free of physical conditions including type 1 or 2 diabetes; uncontrolled hypertension (>140/90 mm Hg); a history of cerebrovascular, cardiovascular, kidney, or liver disease; the use of medications known to affect body weight (e.g., steroids); pregnancy or lactation; and a weight loss ≥5% of initial weight and/or the use of anorectic agents in the previous 6 months. Psychosocial contraindications included bulimia nervosa, major depression, or other psychiatric illness that significantly disrupted daily functioning. Participants also were required to have daily access to the Internet.
Candidates were screened by telephone. Those who remained eligible after the screening were scheduled for a 1-hour interview with a clinical psychologist during which their psychosocial status was assessed, weight and dieting histories were reviewed, study details were described, and informed consent was obtained. Applicants were referred to their family physician for medical evaluation to verify that they were free of contraindications to participation.
This study was a 1-year randomized controlled trial conducted from February 2001 to September 2002. Eligible applicants were randomly assigned to: 1) eDiets.com, a commercial Internet weight loss program (N = 23); or 2) a weight loss manual (N = 24). Brief assessment visits (i.e., 10–15 minutes) to obtain weight and blood pressure were scheduled for both groups at baseline and weeks 2, 4, 8, 12, 16, 20, 26, 34, 42, and 52. The University of Pennsylvania's Committee on Subjects Involving Human Beings approved the study protocol.
Participants assigned to eDiets.com were given a 1-year membership that was paid for directly by the investigators (L.G.W.). Membership provided a virtual visit with a dietitian. Participants were prescribed a diet of conventional foods designed to match their needs, likes, and lifestyle. Women with a BMI of 27 to 35 kg/m2 were provided meal plans that prescribed ∼1200 to 1300 kcal/d, whereas those with a BMI >35 kg/m2 were given 1300 to 1400 kcal/d. Customized grocery lists were provided to assist participants in purchasing appropriate foods. Social support was provided by methods that included access to on-line meetings that were moderated by a professional, on-line bulletin board support groups, an animated fitness instructor, a 24-h/d help desk staffed by customer service representatives, e-mail reminders about the program and their goals, and bi-weekly diet and fitness e-mail newsletters. If participants wanted additional support, eDiets.com's “find a buddy” program allowed members to e-mail with other members of eDiets.com. Physical activity recommendations were tailored to participants' self-reported levels of cardiovascular endurance and muscular strength (i.e., beginner, intermediate, or advanced).
Participants met with a psychologist (L.G.W.) at baseline and at weeks 8, 16, 26, and 52 for ∼20 min/visit. The first visit reviewed the goals and methods of treatment and included recommendations to log on daily to eDiets.com and to record food intake daily during the first 16 weeks. The four subsequent meetings addressed participants' progress and satisfaction with the program.
Weight Loss Manual
Participants in this group were given a copy of the LEARN Program for Weight Management 2000 (18), a 243-page book that provided 16 step-by-step lessons for modifying eating, activity, and thinking habits. The manual instructed women to consume a 1200- to 1500-kcal/d self-selected diet of conventional table foods. Dietary recommendations were based on the Food Guide Pyramid. Participants were encouraged to keep daily records of their food intake, including the number of calories consumed. (Record keeping was emphasized for the first 16 weeks, as in eDiets.com.) The manual encouraged women to increase their physical activity (by walking up to 30 min/d) and to practice other weight control behaviors (e.g., stimulus control, slowing eating, cognitive restructuring). After 16 weeks, participants were given the Weight Maintenance Survival Guide (19) that reiterated concepts introduced in the LEARN Program (18). Participants in this group met with a psychologist on the same schedule (i.e., at baseline and four times over the year) as those in the eDiets.com group.
The primary outcome measure was change in body weight, which was measured on a digital scale at brief assessment visits at baseline and weeks 2, 4, 8, 12, 16, 20, 26, 34, 42, and 52. Research assistants conducted visits. Weight was measured (without shoes) to the nearest 0.1 kg. Height was determined (at baseline) to the nearest 0.1 cm using a wall-mounted stadiometer. Blood pressure was measured at each assessment using a NAMAP monitor (model 9B00J2209; NAMAP, Tampa, FL). A fasting biochemical profile that included triglycerides, glucose, total cholesterol, high-density lipoprotein (HDL),1 and low-density lipoprotein (LDL) was obtained at baseline and weeks 16 and 52.
Mood was assessed by the Beck Depression Inventory-II (20), quality of life was evaluated by the Medical Outcomes Study, Short Form-36 Health Survey (21), and aspects of eating behavior—including restraint, disinhibition, and hunger—were assessed by the Eating Inventory (22). Questionnaires were administered at baseline and weeks 16 and 52.
Participants' behavioral adherence during the first 16 weeks was assessed by completion of food diaries, which were collected at each visit. As in our previous studies, we counted only the number of days per week that participants completed their food records (23,24). Thus, for each week, they received a score of 0 to 7, based on the number of days recorded. Participants in the eDiets.com group also were asked to record each time they logged on to the web site. They were provided a calendar each month on which to record this information.
To help cover travel costs incurred during the study, participants were paid $50 for completing the 52-week assessment.
We hypothesized that participants assigned to eDiets.com would lose a mean of 6.0 ± 3.0 kg after 16 weeks compared with a loss of 3.0 ± 3.0 kg for women in the manual group. We thought that the extensive social support available from eDiets.com would facilitate greater weight loss than the manual-based approach. Using an α of 0.05 and power of 80%, a sample size of 18 for each group was needed to detect a 3.0-kg difference among groups. Assuming a 20% attrition rate, a sample of 45 subjects was selected.
Student's t tests for independent samples were used to assess differences between groups in baseline characteristics. Changes in weight and other outcomes were assessed by repeated-measures ANOVA. All analyses were performed using the Statistical Package for the Social Sciences (SPSS for Windows, version 11.5; SPSS, Chicago, IL). Three analyses were performed on the weight change data: 1) a last-observation-carried-forward (LOCF) analysis that included all randomized participants (in which missing values were imputed by carrying forward the last measured observation); 2) a baseline-carried-forward (BCF) analysis that included all randomized participants (with missing values at weeks 16 and 52 replaced by the participant's baseline value); and 3) a completers analysis that included only participants who completed the week 52 visit (N = 31).
- Top of page
- Research Methods and Procedures
Baseline Characteristics and Attrition
At baseline, there were no significant differences between groups in age, weight, height, or BMI (see Table 1). Attrition was 34% at both weeks 16 and 52 and did not vary by treatment group at either time. Participants who dropped out did not differ significantly from treatment completers on baseline measures of age, weight, height, or BMI. One participant had impaired glucose tolerance at baseline (fasting plasma glucose = 117 mg/dL) and developed type 2 diabetes by week 8 (fasting plasma glucose = 132 mg/dL). This participant, who was in the manual group, was excluded from all outcome analyses.
|Variables||eDiets.com (n = 23)||Weight loss manual (n = 24)|
|Age (years)||44.2 ± 9.3||43.3 ± 11.1|
|Height (cm)||165.5 ± 6.5||162.8 ± 6.3|
|Weight (kg)||93.4 ± 12.6||87.9 ± 10.8|
|BMI (kg/m2)||33.9 ± 3.2||33.0 ± 3.0|
|Blood pressure (mm Hg)|
|Systolic||121.7 ± 16.7||112.1 ± 13.8|
|Diastolic||74.4 ± 10.1||66.0 ± 9.6|
|Glucose (mg/dL)||90.2 ± 11.7||81.5 ± 21.3|
|Total||206.2 ± 37.5||202.1 ± 32.3|
|HDL||53.8 ± 10.6||60.5 ± 12.1|
|LDL||128.0 ± 33.7||118.9 ± 30.4|
|Triglycerides (mg/dL)||122.1 ± 62.1||114.0 ± 62.1|
LEARN participants attended a mean of 8.1 ± 3.4 scheduled visits, whereas those in eDiets.com attended 7.6 ± 3.2 visits. The number of visits attended did not differ significantly between groups. However, collapsing across groups, participants' attendance did decrease significantly over the course of the study (p < 0.001). During the first 16 weeks, participants attended a mean of 82.3 ± 22.1% of their six scheduled visits. In contrast, from weeks 17 to 52, they attended only 58.7 ± 43.1% of the five scheduled visits.
At week 16, participants in eDiets.com lost a mean of 0.9 ± 3.2% of initial weight (i.e., 0.7 ± 2.7 kg) compared with a loss of 3.6 ± 4.0% (3.0 ± 3.1 kg) for those in the manual group (see Figure 2). At week 52, losses for the two groups were 1.1 ± 4.0% (0.8 ± 3.6 kg) and 4.0 ± 5.1% (3.3 ± 4.1 kg), respectively. A repeated measures ANOVA, using LOCF, revealed a significant treatment-by-time interaction for percentage reduction in initial weight (p = 0.02). Participants treated by eDiets.com lost significantly less weight at week 16 (p = 0.01) and at week 52 (p = 0.04) than those who received the weight loss manual. Similar statistical conclusions were reached when the data were analyzed using number of kilograms lost rather than percentage reduction in weight.
Identical analyses were conducted using the BCF (N = 46) and completers data (N = 31). The pattern of results for both sets of analyses was similar to that observed with the LOCF analysis. However, differences between groups did not reach statistical significance. For the BCF repeated measures ANOVA that used percentage reduction in initial weight, the p value was 0.08 (and that for kilograms lost was 0.07). The p value for the completers analysis for percentage weight loss was 0.14 (and that for kilograms lost was p = 0.13).
Using the LOCF data, partial correlation analysis, controlling for treatment condition, showed that participants who attended more clinic visits lost more weight at week 16 (r = 0.41, p = 0.004) and at week 52 (r = 0.38, p = 0.01). Identical analyses were conducted using the BCF and completers data. The pattern of results for both sets of analyses was similar to that observed with the LOCF analysis. However, the partial correlation between total number of visits attended and week 52 weight loss did not reach statistical significance with the completers data (r = 0.31, p = 0.09).
Participants treated by eDiets.com kept food records a mean of 18.3 ± 21.7 days of a possible 112 days (i.e., 16 weeks) compared with 29.0 ± 35.3 days for participants assigned to the LEARN manual. The difference between groups was not statistically significant. However, partial correlation analysis, controlling for treatment condition, revealed that the number of food records kept by participants (during the first 16 weeks) correlated positively with weight loss at week 16 (r = 0.65, p < 0.001) and at week 52 (r = 0.40, p < 0.006).
During the first 16 weeks, eDiets.com participants reported a mean of 17.7 ± 21.1 log-ons to the program's web site. The number of log-ons did not correlate significantly with weight loss at either week 16 or week 52. A median split analysis, however, revealed that participants in the high log-on group lost 2.3 ± 2.5% of initial weight at week 16 and 2.3 ± 4.3% at week 52, which was significantly (p < 0.02) greater than the weight gain of 0.6 ± 3.1% at week 16 and of 0.2 ± 3.3% at week 52 in the low log-on group. Participants in the high and low log-on groups had a mean of 33.0 ± 18.7 and 1.0 ± 2.2 log-ons, respectively.
Changes in eating behavior were assessed by repeated measures ANOVAs, examining only treatment completers. No significant differences were observed between the two groups on any of the measures. However, within-group analyses revealed significant changes over time. As shown in Table 3, cognitive restraint increased significantly (p < 0.001) from baseline at weeks 16 and 52, whereas (dietary) disinhibition and hunger decreased significantly (both p ≤ 0.001). At week 52, participants also reported significantly (p < 0.004) fewer symptoms of depression, as measured by the Beck Depression Inventory-II.
|Variable||Baseline||Week 16||Week 52|
|Restraint||8.4 ± 3.7||12.7 ± 4.1*||13.3 ± 4.6*|
|Disinhibition||10.2 ± 3.8||8.4 ± 4.0*||8.6 ± 3.9*|
|Hunger||6.4 ± 3.3||5.9 ± 4.0||4.7 ± 3.6*|
|Quality of life|
|Physical||82.5 ± 17.7||85.5 ± 16.3||90.8 ± 13.1*|
|Functioning vitality||50.5 ± 24.5||51.9 ± 21.5||62.9 ± 21.7*|
|Depression||9.1 ± 7.7||8.0 ± 8.4||5.1 ± 5.4*|
|LDL||120.4 ± 32.2||135.5 ± 32.7*||125.1 ± 31.2|
|HDL/total||3.8 ± 1.0||4.1 ± 0.9*||3.8 ± 0.8|
Quality of Life
Changes in quality of life were also assessed by repeated measures ANOVAs in treatment completers. No significant differences were observed between the two groups on any of the measures. However, within-group analyses revealed significant changes over time. At week 52, participants reported significant improvements in physical functioning and vitality (p = 0.25 and 0.003, respectively), as assessed by the Short Form-36.
Blood Pressure, Glucose, and Lipids
Repeated measures ANOVAs similarly revealed no significant differences between the two groups in changes in blood pressure, glucose, lipids, or lipoproteins. Within-group analyses revealed significant changes over time in only two variables, and they were in the direction opposite of that expected (see Table 3). Both LDL and the HDL/total cholesterol ratio increased significantly from baseline to week 16 (p < 0.001 and p = 0.018, respectively). However, both variables returned to baseline levels at week 52.
- Top of page
- Research Methods and Procedures
This study's principal finding was that eDiets.com produced minimal weight loss and was not as effective as a traditional manual-based approach. Participants in eDiets.com lost only 0.9% of initial weight after 16 weeks and only 1.1% at 1 year (determined by the LOCF analysis). In contrast, women who received the weight loss manual lost 3.6% and 4.0% of initial weight after 16 and 52 weeks of treatment, respectively.
Both interventions prescribed similar weight loss techniques that included a reduced calorie diet, increased physical activity, and recommendations to record food intake. A major difference between the interventions was that participants in eDiets.com had unlimited access to social support, including on-line meetings that were moderated by a professional, on-line bulletin board support groups, and a 24-h/d help desk staffed by customer service representatives. Because of these differences, we hypothesized that women in eDiets.com would lose twice as much weight as those who received the weight loss manual.
Two factors, however, seemed to limit the potential benefits of eDiets.com. First, participants apparently made minimal use of the services available to them, as reflected by their logging on to the web site an average of only 17.7 times during the first 16 weeks. Results of a median split analysis revealed that women in the high log-on group (who averaged 33.0 log-ons) lost significantly more weight than those in the low log-on group (mean of 1.0 log-on). Even the high users, however, logged on an average of only twice a week, far less than we had expected they would. Harvey-Berino et al. (25) similarly found that participants in a weight maintenance study were less likely to participate in Internet chat groups than they were to attend group sessions on site. Thus, the fact that Internet weight loss programs are convenient—only a few keystrokes away—does not mean that dieters will use them.
The second factor concerned structure. AlthougheDiets.com and the LEARN manual encouraged similar behaviors—including food monitoring, exercise, and calorie counting—eDiets.com did not seem to be as structured as the step-by-step LEARN approach. The LEARN manual provided 16 weekly behavioral weight control lessons that built on one another. For example, each lesson in the manual reiterated the importance of keeping food records and counting calories. eDiets.com participants had self-guided access to food records on the web site, but the program did not emphasize record keeping and calorie counting in a manner comparable with the regimented approach in LEARN. When recording an entry in the eDiets.com food journal, members were asked to record the food eaten, time, quantity, with whom, and mood. They were not specifically asked to record and total their calories. By contrast, participants who used the LEARN manual understood that keeping daily food and calorie records was an essential part of the program.
Results of two previous studies, conducted in an academic medical center, suggest that Internet weight loss programs are likely to be most effective when they require participants to keep daily records of their food intake and physical activity, as they do in traditional group behavioral programs (8,9). These two Internet programs require participants to submit their records electronically to their counselor, who then provided feedback on their performance. In the first study, participants who received this behavioral e-counseling lost 4.0 kg in 6 months (8), and those in a second investigation lost 4.4 kg at the end of 1 year (9). In both studies, behaviorally treated participants lost more than twice as much weight as persons who received information about appropriate eating and activity but were not required to monitor their behavior in the same manner (8,9). The importance of monitoring food intake is supported by the positive relation observed in the present study between weight loss and the number of days that food records were kept.
Weight loss in this study was associated with increases in cognitive restraint and with decreases in dietary disinhibition and hunger, as reported by other investigators (26,27,28). We similarly observed favorable changes in mood, physical functioning, and vitality, as reported by others (29,30,31). We are unable to explain the increases in LDL cholesterol and the HDL/total cholesterol ratio. Foster et al. (32) have observed small reductions in these values in patients who received the LEARN manual and lost ∼3% of initial weight.
The attrition rate of 34% at both weeks 16 and 52 was higher than we had anticipated. Studies by Tate et al. have reported attrition of 22% after 6 months (8) and 16% after 12 months (9). We cannot explain our high attrition rate but suspect that it was attributable, in part, to the minimal weight loss achieved by participants.
Additional studies, with larger sample sizes (that include both men and women), are needed to assess the efficacy of eDiets.com, as well as other Internet-based weight loss programs. In designing these studies, it may be useful to collaborate with industry representatives to obtain a precise count of the number of times participants log on to the web site and for how long. We did not collaborate with eDiets.com in conducting this study. Investigators also are encouraged to present the Internet interventions in the manner in which the public will use them. In this study, participants were weighed in our clinic at 11 brief visits, and they met with a psychologist on 5 of these occasions. Thus, results of this study likely represent a best-case scenario concerning the efficacy of eDiets.com (as well as the LEARN manual). The 11 brief visits were included to facilitate participants' retention in the study, but they also may have increased weight losses compared with those resulting from the provision of eDiets.com (or the manual) alone.
This study represents, to our knowledge, the first evaluation of a commercially-based Internet weight loss program. As such, it responds to the Federal Trade Commission's call for commercial programs to disclose their results of treatment so that consumers can make informed decisions when selecting a weight loss plan (11,33). These findings suggest that consumers are unlikely to achieve clinically significant weight loss by using Internet programs that provide primarily information about diet and physical activity. Successful Internet interventions seem to incorporate practices—including record keeping, personalized feedback, and accountability—that are found in traditional behavioral weight control programs. It remains to be seen whether commercial Internet programs can incorporate such practices.
- Top of page
- Research Methods and Procedures
This research was supported, in part, by a Pilot Study Grant from the North American Association for the Study of Obesity (to Dr. Womble) and by NIH Grant K24-DK-065, 018 (to Dr. Wadden).
Nonstandard abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein; LOCF, last-observation-carried-forward; BCF, baseline-carried-forward.
- Top of page
- Research Methods and Procedures
- 32002) Behavioral treatment of obesity. In Wadden, TA and Stunkard, AJ (eds). Handbook of Obesity Treatment. Guilford: New York. pp 301–316.(
- 112002) Commercial and self-help weight loss programs. In Wadden, TA and Stunkard, AJ (eds). Handbook of Obesity Treatment. Guilford: New York. pp 395–415., , (
- 12Weight Watchers. (2003) Available online at http:www.weightwatchers.comindex.aspx. Accessed October 22.
- 13Jenny Craig. (2003) Available online at http:www.jennycraig.com. Accessed October 22.
- 14Nutrisystem. (2003) Available online at http:www.nutrisystem.com. Accessed October 22.
- 15Diet Smart. (2003) Available online at http:www.dietsmart.comcgi-bindietsmartIndex.sb. Accessed October 22.
- 17Federal Trade Commission. (1997) Commercial Weight Loss Products and Programs: What Consumers Stand to Gain and Lose. Federal Trade Commission: Washington, DC.
- 182000) The LEARN Program for Weight Management 2000. American Health Publishing: Dallas, TX.(
- 191990) Weight Maintenance Survival Guide. American Health Publishing: Dallas, TX., (
- 201987) Manual for the Beck Depression Inventory. Psychological Corp.: New York., (
- 221988) Eating Inventory Manual. Psychological Corp.: New York., (
- 332003) What consumers want to know about commercial weight loss programs: a pilot investigation. Obes Res. 40: 131–135., , , et al. (