RESUMO
BACKGROUND: The effect of computer- or human-delivered personalized feedback on the effectivess of web-based behavior change platforms for weight loss is unclear. OBJECTIVE: We aimed to compare the effectiveness of a web-based behavior change intervention personalized through either computerized or human-delivered feedback with a nonpersonalized intervention in promoting weight loss in community-based adults with overweight or obesity. METHODS: This pragmatic, 3-group, parallel-arm, randomized trial recruited students and staff in a Brazilian public university who were aged 18 to 60 years, had a BMI of ≥25 kg/m2, and were not pregnant. Participants were allocated to one of 3 groups: platform only (24-week behavior change program delivered using a web platform with personalized computer-delivered feedback), platform plus coaching (same 24-week web-based behavior change program plus 12 weeks of personalized feedback delivered online by a dietitian), or waiting list (nonpersonalized dietary and physical activity recommendations delivered through an e-booklet and videos). Self-reported weight at 24 weeks was the primary outcome. Changes in dietary and physical activity habits within 24 weeks were secondary outcomes. RESULTS: Among the 1298 participants, 375 (28.89%) were lost to follow-up. In the intention-to-treat analysis, the platform-only and platform plus coaching groups had greater mean weight loss than the waiting-list group at 24 weeks (-1.08 kg, 95% CI -1.41 to -0.75 vs -1.57 kg, 95% CI -1.92 to -1.22 vs -0.66 kg, 95% CI -0.98 to -0.34, respectively). The platform-only and platform plus coaching groups, compared with the waiting list group, had a greater increase in the consumption of vegetables (3%, 95% CI 1% to 6% vs 5%, 95% CI 2% to 8% vs -3%, 95% CI -5% to 0%) and fruits (9%, 95% CI 6% to 12% vs 6%, 95% CI 2% to 9% vs 2%, 95% CI 0% to 6%) and a larger reduction in ultraprocessed food intake (-18%, 95% CI -23% to -13% vs -25%, 95% CI -30% to -20% vs -12%, 95% CI -16% to -8%). Changes in physical activity did not differ across the groups. Engagement was higher in the platform plus coaching group than in the platform-only group (7.6 vs 5.2 completed sessions; P=.007). Longer usage of the platform was associated with clinically meaningful (≥5%) weight loss (odds ratio 1.02, 95% CI 1.01 to 1.04). CONCLUSIONS: The web-based behavior change programs with computer- and human-delivered personalized feedback led to greater, albeit small-magnitude, weight loss within 24 weeks. Improvement in multiple dietary habits, but not physical activity, were also greater in the personalized programs compared with the nonpersonalized one. The human-delivered personalized feedback by the online dietitian coach increased user engagement with the program and was associated with a significantly higher chance of clinically meaningful weight loss. TRIAL REGISTRATION: ClinicalTrials.gov NCT03435445; https://clinicaltrials.gov/ct2/show/NCT03435445. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.2196/10.1186/s12889-018-5882-y.