The reviewed studies suggest that internet-based interventions do not have a uniform effect on weight. Using intent to treat (ITT) methods, weight loss ranged from less than 1 kg (considered to be natural noise) to 4.9 kg. Since interventions differed considerably, results are presented individually by study. Statistical pooling was not employed due to the differences between studies in interventions. This review was conducted with a primary interest in the efficacy of internet interventions for weight loss and the magnitude of weight change. Notably, participant attrition surfaced as an issue because it attenuates weight loss results when using ITT methods and raises concern about the feasibility of promoting interventions on a larger scale. Participant characteristics and weight loss are presented in . summarizes the percent weight loss from baseline to study completion based on ITT analysis.
Clinical Relevance of Weight Change Among Internet Groups at Study Completion
In the Tate et al. study 91 participants were randomized to either internet education or internet behavioral therapy. All participants were offered a website orientation and were given weight loss resources and behavioral weight control lessons13
. This initial lesson included advice on calorie intake and physical activity and explained self-monitoring tools and online resources. Self-monitoring was encouraged for all participants, but only the behavioral therapy group submitted weekly self-monitoring diaries to the therapist. All participants were contacted at 3 and 6 months for follow up measurements and a clinical psychologist visit. Behavioral therapy group participants received weekly emails with structured weight loss lessons and individualized feedback, and were able to submit questions to the therapist. Those who did not send a weekly log received a personalized email inquiring about progress and encouraging continued participation. Among behavioral therapy completers, average weight loss was 4.0 ± 2.8 kg at 3 months and 4.1 ± 4.5 kg at 6 months. Among the education group completers, average weight loss was 1.7 ± 2.7 kg at 3-months and 1.6 ± 3.3 kg at 6 months. This study shows significant differences between groups at both time points (p=.001, p=.04 respectively). ITT analysis also showed a statistically significant difference between groups, in addition to a significant treatment x
time interaction. However, ITT weight loss among the internet group at 6 months showed a loss of only 2.9 ± 4.5 kg at 6 months. Login frequency was correlated with weight change between 0 and 6 months for behavioral therapy (r=−0.43, p=.003) and the education only group (r=−0.33, p=.03). Attrition did not vary by treatment group at either 3 or 6 months, but those who did not attend the 6 month visit were younger and had less email or Internet experience at baseline. This study shows that the personalized attention through the behavioral therapy intervention had a greater effect on weight loss than the generic internet-based education. Still, the internet group reduced body weight by only 3.7%, less than a clinically relevant weight change.
Womble et al. offered all 47 participants baseline and quarterly meetings with a psychologist, but randomized participants either to support through an online resource (eDiets) or a weight loss manual (LEARN).14
Researchers believed that constant and continuous access to the website, in combination with a virtual visit with a dietitian, customized grocery lists, physical activity recommendations, and unlimited access to social support would lead to greater weight loss than the LEARN manual. The eDiets group lost only 0.8 ± 3.6 kg at 52 weeks, compared to the LEARN weight loss of 3.3 ± 4.1 kg (p=.04). Researchers found that participants used the website less than anticipated, averaging 17.7 log-ins during the first 16 weeks. Employing a median split for log-in frequency, a positive association was found with weight loss among the eDiets group. A lack of power may in part explain the null results as 34% attrition was greater than anticipated. Analysis showed that although participant attendance did not significantly differ by intervention group, attendance decreased significantly over time. The average weight loss of 0.9% among the internet intervention group is not clinically relevant.
Gold et al. compared eDiets, the commercial self-help website, with VTrim, an online therapist-led structured behavioral weight loss program among 124 individuals. VTrim entailed a 6 month weight loss phase and a 6 month weight maintenance phase15
. The VTrim intervention focused on weekly behavioral and self-management strategies. Self-reported weight was discussed weekly and feedback was offered on completed assignments. In the maintenance phase feedback was offered bi-weekly and participants were encouraged to use support from the website. The eDiets group self-reported weight weekly and received automated feedback, as well as individually tailored meal plans and exercise programs. Although the eDiets website did not offer a structured behavioral plan with lessons or activities, it hosted online meetings, chat rooms, discussion boards and buddy systems. VTrim was more effective than eDiets in promoting weight loss among completers (8.3 ± 7.9 kg vs. 4.1 ± 6.2 kg at 6 months; 7.8 ± 7.5 kg vs. 3.4 ± 5.8 kg at 12 months). Baseline carried forward (BCF) analysis showed attenuated results (5.1 ± 7.1 kg vs. 2.6 ± 5.3 kg at 12 months). Attrition for VTrim was 18% at 6 months and 35% at 12 months, while eDiets attrition was 19% at 6 months and 23% at 12 months. This difference was not significant by treatment group, although within group analysis showed that VTrim completers were more highly educated than VTrim dropouts. Adjustment for log-in frequency eliminated weight loss differences between groups, showing both groups averaging a weight loss of 6.2 ± 1.0 kg (p=0.996 for between group differences). This study suggests that VTrim behavioral strategies were more effective than eDiets in inducing weight loss, but log-in frequency explains the difference in weight loss between groups. The eDiets group on average lost 2.9% of baseline body weight, falling short of the 5% clinically relevant standard.
The McConnon et al. study randomized 221 participants to either a website intervention or usual care. The website group was offered targeted advice, tools and information.16
Self-reported weight loss progress was stored online and automatic generic emails encouraging website usage were generated if participants did not log-in weekly. Those randomized to usual care were asked to continue with usual approaches to weight loss and were offered print information at baseline in accordance with information given in primary care. Researchers recorded a weight loss of 1.3 kg in the website-based group and 1.9 kg in the usual care group, showing no statistically significant difference between groups for weight loss at 12 months. The low change in weight may be due to extensive attrition, and also reflects the low intensity of the intervention. Since researchers anticipated and recruited to account for 22% attrition, the 41% attrition observed greatly affects the power of the study to detect a weight loss difference. Researchers therefore used both last observation carried forward (LOCF) and BCF in the analysis, biasing results toward the null. Adjusting for age, sex, baseline weight, physical activity and baseline confidence, the internet group was heavier at 12 months than at baseline (0.5 to 0.8 kg). This study shows that internet-based weight control package developed for this study and offered to obese volunteers was not effective in causing weight loss. Average body weight was reduced by 1.3% among the internet intervention group, less than the clinically relevant goal.
The Rothert et al. study compared weight loss between an online tailored information group and an online information only group (n=2,862).17
The tailored intervention, a weight management program called Balance, was a 6 week self-help program offering tailored plans formulated by a software algorithm using baseline survey responses. Baseline guides were supplemented by tailored action plans at weeks 1, 3 and 6 delivered via email notification. The resulting 2.8 ± 0.3 kg weight loss among tailored expert system completers at 6 months is significantly greater than the 1.1 ± 0.4 kg lost by the information only group (p= 0.0007), but is slightly less in magnitude than the commercial website results seen by Gold et al. Analysis using measurements at 3 and 6 months was performed both for completers and using LOCF, showing great attenuation using LOCF. There were no significant differences in response rates between the treatment groups, although those above age 55 and whites were most likely to respond compared with African Americans and Hispanics. Still, it is notable that 36% of study participants were African American, a population not well represented in previous internet-based weight loss studies. Researchers conducted a telephone survey of non-respondents to inquire about current weight and learned that of the 27 individuals interviewed, 11 reported not receiving email notification of the follow-up surveys. This study shows that an automated web-based intervention is more effective than information in leading to weight loss, although communication difficulties and ethnic differential responses present directions for further investigation. Body weight was reduced by an average of 3.0% among the internet intervention group, less than the clinically relevant standard.
The Behavioral Intervention Therapy delivered through the Hunter et al. study incorporated aspects of the previously discussed interventions, but at a higher intensity dose. The 446 participants were active-duty military personnel recruited through a military medical research center. After attending in-person orientation, participants were offered behavioral, dietary and exercise plans18
. Researchers instructed participants to log-in 5 times per week to complete self-monitoring food and exercise diaries. Weekly lessons included behavioral modification strategies and interactive quizzes lasting 20–30 minutes. Participants also received the LEARN Program for Weight Management Manual and were assigned weekly reading. In addition, participants received 15-minute motivational interviewing telephone calls at weeks 4 and 8. Feedback and phone calls were offered by a single counselor. Researchers reported a 1.3 ± 4.1 kg weight loss among the intervention group and a weight gain of 0.6 ± 3.4 kg among the usual care group at 6 months using BCF. Although there was only a small difference between groups for weight loss, researchers noted a significant positive association between weight change and site use as measured by logins or self-report. Furthermore, 22.6% of intervention participants lost 5% or more of body weight as compared to only 6.8% of usual care participants. Almost 83% of the sample was evaluated at 6 months, despite the fact that there was no run-in phase or incentives for participation. Researchers did not find any significant interactions between treatment group assignment and BMI, race or gender. This study demonstrates the need for a weight loss intervention that is more effective among this young and ethnically diverse study population. The reported 1.5% average reduction in body weight does not meet the goal for clinical relevance.
Bennett et al. (2009) compared usual care (individual providers offering advice as they liked and “Aim for a Healthy Weight” print materials) with an interactive web-based intervention among 101 individuals.19
The intervention group was offered tailored obesogenic behavior change goals. Intervention participants worked with a health coach to establish goals. Given earlier evidence that adherence wanes over time, researchers created a dynamic self-monitoring tool comparing individual progress with average changes of other participants and reassessed goals at 6 weeks. Participants were encouraged to use the website at least 3 times per week, earning a raffle ticket with each log-in and/or tracked behavior so as to create further incentive for participation. A coach was available via messaging, and social networking forums and recipes were offered online. In addition, a health coach initiated two, 20-minute motivation-coaching sessions in person at baseline and 6 weeks and two 20-minute biweekly sessions over the phone at weeks 3 and 9. Researchers found a greater between group difference in weight loss analyzing the completers (−3.05 kg) than using BCF (−2.56 kg). Total mean weight loss among the completers was 2.71 ± 3.34 kg among the intervention group and 0.34 ± 2.04 kg gain among the control group. Of clinical relevance is the finding that 25.6% of intervention participants and no control participants lost >5% of initial body weight at 12 weeks. While more men than women were lost to 3 month follow up, other demographic differences between completers and non-completers were not statistically significant. This study shows a statistically significant difference in weight loss between the intervention and control groups and good adherence to log-in goals, but the magnitude of weight loss is less than comparable trials. The 2.3% average body weight reduction reported is less than the clinically relevant standard.
The Harvey-Berino et al. intervention differed from the above reviewed studies in that it included an intensive 6 month interactive TV-delivered component and then utilized the internet for a weight maintenance program among 255 participants. This interactive portion of the intervention entailed weekly one hour discussions, print outs, instructions on calorie reduction, and encouragement of physical activity, in addition to dietary and physical activity logs20
. After the intervention, weight maintenance was tracked through 3 support arms: Internet-based (IS), minimal in-person (M-IPS) and frequent in-person (F-IPS). Final measurements of weight maintenance from baseline were greater in the IS group (−4.7± 6.9 kg) than in the in-person support groups (−3.9± 5.9 kg for M-IPS, and −4.2± 7.9 kg for F-IPS). The reported ITT analysis attenuates weight loss results, and completer analysis showed a greater magnitude of weight loss as shown in . Higher attrition rates in the IS group were associated with less education and less weight loss during the first 6 months of treatment. The clinically relevant weight loss and maintenance of 5.2% of body weight shows promise for internet maintenance interventions and should be evaluated in future studies.
Like the previously discussed study, Cussler et al. offered weight loss treatment followed by an internet delivered weight maintenance program. The 4 month in-person treatment entailed a weekly session to focus on physical activity, nutrition and healthy eating, social support and mind/body wellness skills. After 4 months, participants were cluster randomized to either internet-based or self-directed weight maintenance. The internet group had a mean weight loss of 5.3 ± 3.6 kg, while the self-directed group had a mean loss of 5.2 ± 3.8 kg. Over the following 12 months participants in the internet-based group had access to tools for communication, progress monitoring logs, curriculum materials including dietary and activity information and links to relevant websites. Using BCF analysis, the difference in weight maintenance between groups were not statistically significant. Women in the internet group regained 0.4 ± 5.0 kg compared to a regain of 0.6 ± 4.0 kg in the control group. Authors noted a higher attrition rate of 21.2% in the internet group than the 14.5% attrition in the control group, citing discomfort with the internet as a possible explanation. Also, self-directed participants organized groups with peers from the weight treatment sessions to compete with the internet group (the “Avis” effect), which may bias results toward the null. Researchers tried to isolate the type of internet use by associating logs of diet, weekly weight, physical activity, and frequency of contact with 4–16 month weight change. Diet log completers were significantly and negatively correlated with change in body weight (rho= −0.29, p<0.05). Changes in exercise energy expenditure were positively correlated with multiple internet log entries (diet log, rho= 0.41; weekly weight, rho = 0.41; physical activity, rho=0.33; total, rho=0.38, p<0.01). While the internet group maintained a clinically relevant 5.8% body weight loss to study completion, there was no statistically significant difference between groups, calling into question the specific impact of the internet intervention.