This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
There is a paucity of information in the scientific literature on the effectiveness of commercial weight loss programs, including Web-based programs. The potential of Web-based weight loss programs has been acknowledged, but their ability to achieve significant weight loss has not been proven.
The objectives were to evaluate the weight change achieved within a large cohort of individuals enrolled in a commercial Web-based weight loss program for 12 or 52 weeks and to describe participants’ program use in relation to weight change.
Participants enrolled in an Australian commercial Web-based weight loss program from August 15, 2007, through May 31, 2008. Self-reported weekly weight records were used to determine weight change after 12- and 52-week subscriptions. The primary analysis estimated weight change using generalized linear mixed models (GLMMs) for all participants who subscribed for 12 weeks and also for those who subscribed for 52 weeks. A sensitivity analysis was conducted using the last observation carried forward (LOCF) method. Website use (ie, the number of days participants logged on, made food or exercise entries to the Web-based diary, or posted to the discussion forum) was described from program enrollment to 12 and 52 weeks, and differences in website use by percentage weight change category were tested using Kruskal-Wallis test for equality of populations.
Participants (n = 9599) had a mean (standard deviation [SD]) age of 35.7 (9.5) years and were predominantly female (86% or 8279/9599) and obese (61% or 5866/9599). Results from the primary GLMM analysis including all enrollees found the mean percentage weight change was −6.2% among 12-week subscribers (n = 6943) and −6.9% among 52-week subscribers (n = 2656). Sensitivity analysis using LOCF revealed an average weight change of −3.0% and −3.5% after 12 and 52 weeks respectively. The use of all website features increased significantly (
The weight loss achieved by 12- and 52-week subscribers of a commercial Web-based weight loss program is likely to be in the range of the primary and sensitivity analysis results. While this suggests that, on average, clinically important weight loss may be achieved, further research is required to evaluate the efficacy of this commercial Web-based weight loss program prospectively using objective measures. The potential association between greater website use and increased weight loss also requires further evaluation, as strategies to improve participants’ use of Web-based program features may be required.
As the prevalence of overweight and obesity among adults continues to increase across the world [
The most recent systematic review of major commercial weight loss programs concluded that there was inadequate evidence to recommend their use [
Commercial weight loss program providers commonly offer Web-based versions of their programs. Recent systematic reviews of Web-based weight loss interventions have highlighted the potential of these programs to achieve significant weight loss [
A 2010 systematic review of Web-based weight loss interventions found that greater weight loss is likely to be associated with increased use of Web-based program features [
Therefore, the primary aim of this study was to describe the weight loss achieved by a cohort of enrollees of a commercial Web-based weight loss program among participants who subscribed to the program for 12 or 52 weeks. The secondary aim was to describe participants’ use of the Web-based program overall and by percentage weight loss category and to determine if website use differed by percentage weight loss category.
Participants were eligible for inclusion in the study if they paid for a subscription to the program from August 15, 2007, through May 31, 2008. To join the program, participants must have been 18 to 75 years of age and have had a body mass index (BMI) greater than or equal to 22 kg/m2 based on self-reported height and weight. When participants enrolled, they purchased a subscription plan of 4-, 12-, 16- or 52-weeks duration. In 2007-2008, a subscription cost A$16.50 to A$79.95 per month dependant on the number of months a participant subscribed. Participants could not unsubscribe from their selected plan until the subscription timeframe had elapsed unless they had special circumstances that prevented them from completing their subscription (eg, pregnancy or financial difficulties). This study included participants who subscribed for the most popular durations of 12- or 52-weeks. Data related to free or non-consecutive subscriptions (≥ 7 days apart) were also excluded.
Characteristics of the full cohort [
In 2007-2008, SP Health Co Pty Ltd (Sydney, Australia) offered a Web-based weight loss platform that was commercially available in Australia as The Biggest Loser Club (www.biggestloserclub.com.au). It was promoted as a 12-week program, but participants could choose to subscribe for longer to assist with further weight loss and/or maintenance. The self-directed program incorporated evidence-based weight management strategies and aligned with key elements of social cognitive theory [
All data were collected by SP Health Co, provided to the researcher in deidentified form, and included enrollment survey responses (anthropometric measures, ie, weight and height, and demographics, ie, age, gender, and postcode), subscription data (date of enrollment, date membership ceased, and subscription plans held), website use (date of log-in, online food and exercise diary entries, and posts to the discussion forum), and self-reported weight records (date of record and weight recorded). Ethics approval for the study was obtained from the University of Newcastle Human Research Ethics Committee.
Participants’ characteristics were captured from the enrollment survey. Self-reported height and weight were used to calculate BMI (weight in kg/height in m2), which was categorized as healthy, overweight, or obese using the World Health Organization’s (WHO) BMI classification [
Data relating to the subscription plans participants held were used to determine whether participants enrolled for 12 or 52 weeks. The date of enrollment and the date that membership ceased were used to calculate the number of days each participant was a member of the program and, therefore, how many participants cancelled their subscription. The self-reported weight records were used to describe the number of people who weighed in each week. The self-reported weights (in kilograms) were used to determine the weight change achieved. The total number of days per week each of the website features (log-ins, food diary entries, exercise diary entries, and forum posts) were used was calculated to describe overall website use.
Data analysis was undertaken using Stata 11.0 (StataCorp, College Station, Texas, USA), with
Absolute and percentage weight change were calculated from enrollment to 12 weeks for participants who subscribed for 12 weeks and from enrollment to 52 weeks for participants who subscribed for 52 weeks. The primary analysis, to determine the weight change achieved by all program enrollees, was conducted using generalized linear mixed models (GLMMs) containing available self-reported weight records for all participants. GLMM was used because this is the preferred method for longitudinal data with missing values [
A secondary sensitivity analysis was conducted to determine the robustness of the results from the GLMM approach. This analysis was required as GLMM are based on the assumption that missing data are missing at random, which many not be the case for data reported as part of a weight loss program. Therefore, a sensitivity analysis was conducted by imputing missing data for weight using the last observation carried forward (LOCF) method.
Spearman’s rank correlations were calculated to explore associations of weight change with website use. This included the percentage weight change results from the LOCF analyses. Participants were divided into four percentage weight loss categories (weight gain, 0% to < 5% weight loss, 5% to < 10% weight loss, and ≥ 10% weight loss) based on the LOCF analysis results. The median and IQR website use was described by percentage weight loss group and differences between groups investigated using Kruskal-Wallis test for equality of populations.
Participant flow is reported in
Participant flow through the trial.
The proportion of participants who self-reported their weight each week declined substantially over time (
Percent of participants who weighed in per week for 12- and 52-week subscribers.
Weight change results for 12- and 52-week subscribers are shown in
The sensitivity analysis using LOCF gave a mean self-reported weight loss of −2.6 kg (95% CI −2.7 kg to −2.5 kg) or −3.0%, and 21% (1479/6943) achieved greater than or equal to 5% weight loss after 12 weeks (
Mean (95% CI) weight change for a cohort of participants who subscribed to a commercial Web-based weight loss program for 12 or 52 weeks using GLMM and LOCF analyses
Cohort and Weight Change Measure | GLMM Analysisa,b | LOCF Analysisa,b | ||
|
||||
Absolute weight change (95% CI) | −5.6 kg (−5.8 kg to −5.5 kg) | −2.6 kg (−2.7 kg to −2.5 kg) | ||
Percentage weight change (95% CI) | −6.2% (−6.3% to −6.1%) | −3.0% (−3.0% to −2.9%) | ||
|
||||
Weight gain, n (%) | 423 (6.1%) | |||
0% to < 5%, n (%) | 5041 (72.6%) | |||
5% to < 10%, n (%) | 1206 (17.4%) | |||
10% or more, n (%) | 273 (3.9%) | |||
|
||||
Absolute weight change (95% CI) | −8.4 kg (−9.0 kg to −7.8 kg) | −3.6 kg (−3.8 kg to −3.3 kg) | ||
Percentage weight change (95% CI) | −6.9% (−7.3% to −6.5%) | −3.5% (−3.8% to −3.3%) | ||
|
||||
Weight gain, n (%) | 424 (16.0%) | |||
0% to < 5%, n (%) | 1455 (54.8%) | |||
5% to < 10%, n (%) | 475 (17.9%) | |||
10% or more, n (%) | 302 (11.4%) |
aDifference from baseline to 12 and 52 weeks is statistically significant for all analyses (
bControlled for baseline age, BMI, socioeconomic status, and remoteness
Website use for 12- and 52-week subscribers is presented in
Description of 12- and 52-week subscribers’ use of the website features
12-Week Subscribers (n = 6943) | 52-Week Subscribers (n = 2656) | |||
Participants Who Used the Feature, n (%) | Median (IQR) | Participants Who Used the Feature, n (%) | Median (IQR) | |
Log-ins | 6682 (96.2%) | 13 (6-26) | 2576 (97.0%) | 21 (7-56) |
Food diary entries | 5244 (75.5%) | 7 (1-20) | 1993 (75.0%) | 8 (1-34) |
Exercise diary entries | 4686 (67.5%) | 3 (0-9) | 1801 (67.8%) | 3 (0-15) |
Posts to the discussion forum | 860 (12.4%) | 0 (0-0) | 1055 (39.7%) | 0 (0-0) |
For both 12- and 52-week subscribers, percentage weight change was significantly positively correlated (P < .001) with the number of days each website feature was used (
Spearman correlations between website use and percentage weight change (kg) among 12- and 52-week subscribers
12-Week Subscribers (n = 6943) |
52-Week Subscribers (n = 2656) |
|
Log-ins | −0.55 | −0.43 |
Food diary entries | −0.39 | −0.33 |
Exercise diary entries | −0.38 | −0.33 |
Forum posts | −0.12 | −0.18 |
aAll are statistically significant (
The median number of days participants used each website feature increased significantly (
Median (IQR) days each website feature was used by 12- and 52-week subscribers by categories of percentage weight change.
The primary aim of this paper was to describe the weight loss achieved by a large cohort of participants who subscribed to a commercial Web-based weight loss program for either 12 or 52 weeks. The study addresses an existing gap in the literature [
Our primary analysis using GLMM indicated that both 12- and 52-week subscribers achieved statistically significant weight loss. Mean weight loss also exceeded the benchmark (≥ 5%) for clinically important weight loss and improvement in weight-related morbidity, particularly incidence of type 2 diabetes mellitus [
However, the sensitivity analysis at both time points demonstrated less weight loss compared with the GLMM. GLMM assumes that any data missing from the model follow the same trajectory as the included data (in this case weekly weight change). As the average number of weekly weight records included was low and most people self-reported their weekly weight within the initial weeks of the program only, the GLMM results may be biased toward those who self-reported more weekly weights. It is likely that the participants who did not enter their weights were the less successful participants. This is supported by our previous findings that participants with poor eating and activity habits were more likely to stop using the program [
Results from the only two RCTs conducted using another commercial Web-based weight loss program, eDiets, reported a mean percentage weight change of −2.8% [
The second aim of the paper was to describe participants’ use of the Web-based program and its features and to determine if website use was associated with degree of weight loss.
The study demonstrated a significant positive correlation between the number of times each website feature was used and weight change. Therefore, the results support previous research [
However, at the group level, the average use of the commercial Web-based weight management program features appears to be low and inconsistent. The majority of subscribers log on and try the Web-based diary at least once; however, engagement decreases quite fast. This is demonstrated by the initial decline in weekly self-reported weight records over time for both 12- and 52-week subscribers and is consistent with other public health interventions delivered via the Internet, where usage declines after the initial weeks of the intervention [
As this commercial Web-based weight loss program is self-directed, the intensity or frequency of website use is not prescribed. Therefore, this study provides valuable data and insight into what level of website use may be feasible and, more importantly, what level is required to be effective in achieving weight change in a commercial setting. Interestingly, participants who achieved significant weight loss did not use the website unrealistically or excessively. For example, those who achieved greater than or equal to 10% weight loss from baseline to 12 weeks logged on approximately 40% of the possible days (34 days out of 84) and used the Web-based diary 30% of possible days (25 days out of 84). These findings suggest that developing program targets for weekly or monthly website use and for specific program features may increase usage and enhance weight loss, thus facilitating achievement of participants’ weight loss goals. However, to identify optimal exposure to the website overall, as well as individual website features, further investigation of the differences in use at different stages of the program and its association with weight loss is required. For example, this study demonstrates that participants who achieved greater than or equal to 10% weight loss from baseline to 12 weeks logged on approximately 40% of the possible days (34 days out of 84), whereas those who achieved the same percentage weight change from baseline to 52 weeks logged on approximately 22% of the days (81/365). Therefore, further research is needed to investigate the relationship between patterns of website use over time and the weight loss achieved at different time points.
There are several important considerations when interpreting the weight change results. First, the weight change results are based on self-report, and weight is commonly underreported [
The website use data and the reported associations with weight change also have some limitations to be noted. First, the study did not consider use of all website features as these data were not available at the time of the study. Additional data concerning the use of all features (eg, weekly tutorials), as well as more detailed data on the reported features (eg, whether participants read the forum posts) would help to better understand participants’ engagement with the website and the relationship between weight loss and website use. Second, the analysis to determine if greater website use was associated with enhanced weight loss relied on the results of the LOCF analysis. As previously stated, the true weight loss achieved by all participants is likely to be somewhere in the range between the GLMM and LOCF results. Third, although an association between website use and weight loss was demonstrated, a large number of other factors may have influenced participants’ website use and/or weight loss (eg, self motivation, intention to change, and other weight loss strategies) that were not evaluated in this study. Therefore, the association between website use and weight loss must also be confirmed prospectively in an objective manner.
In summary, this research provides important data on an underevaluated weight loss program medium in a large number of commercial program users. The weight loss achieved by 12- and 52-week subscribers of a commercial Web-based weight loss program is likely to be in the range of the primary and sensitivity analysis results. This suggests that, on average, clinically important weight loss may be achieved. The findings support the need for further research to evaluate the efficacy of Web-based weight loss programs and to assist in the development of strategies to increase participants’ ongoing use of Web-based program features.
We acknowledge the work of Ben Noblet in retrieving the data, Anna Crook and Penelope Jones from SP Health for assistance with managing the data set, and Patrick McElduff for assistance with design and implementation of the statistical analyses. MJ Neve was funded by an Australian Postgraduate Award scholarship and a scholarship top-up from SP Health Co Pty Ltd. CE Collins is supported by an Australian National Health and Medical Research Council Career Development Award research fellowship.
MJ Neve received a postgraduate scholarship top-up from SP Health Co. CE Collins has been a consultant dietitian to SP Health Co. PJ Morgan and CE Collins hold an Australian Research Council (ARC) Linkage project grant that is evaluating a weight loss program with SP Health Co.
Accessibility/Remoteness Index of Australia
body mass index
confidence interval
generalized linear mixed model
Index of Relative Socioeconomic Advantage and Disadvantage
interquartile range
last observation carried forward