Wegovy users reported lower food-noise scores in two new telehealth studies, offering a closer look at an experience that weight measurements alone cannot capture. The important qualification is who was counted: people who stayed on treatment and supplied follow-up information.
In its November 14 announcement at ObesityWeek 2026, Novo Nordisk described SIGNAL, an injection study, and OCTANE, a pill study, conducted with the Ro platform. Both involved adults with overweight or obesity and excluded people with diabetes. These are sponsor-reported observational findings, not a randomized comparison or a new approval.
Food noise means more than thinking about dinner
The 2025 paper defining food noise describes persistent food-related thoughts that feel unwanted or distressing and may interfere with daily life. Planning a meal, enjoying food or noticing hunger does not automatically fit that description. The distinction matters when a familiar online phrase becomes a research outcome.
The new studies used the seven-item Ro Allison Indiana Dhurandhar–Food Noise Inventory, or RAID-FN, according to the sponsor. Items are rated from 1 to 5, with higher ratings indicating greater intensity. The 2025 definition paper describes early questionnaire development; it does not itself establish the final seven-item instrument’s validation results.
What changed in the injection study
SIGNAL included 26,213 adults starting semaglutide injection through Ro between May 2025 and June 2026. Participants had not previously used GLP-1 medicines or other weight-management medications. They needed baseline food-noise and weight information, at least one monthly assessment, and continued treatment without a gap longer than 30 days.
The sponsor’s SIGNAL results give an adjusted baseline RAID-FN score of 2.72. The adjusted mean change was −0.42 points at one month and −0.72 points at six months. Researchers used a repeated-measures model accounting for baseline score, demographic and clinical characteristics.
That does not mean all 26,213 people supplied a six-month measurement. The announcement does not provide a month-by-month table of observed participant numbers. A modeled average among an eligible cohort answers a narrower question than what everyone starting treatment can expect, including those who stop early.
Why “25% lower” needs a careful translation
OCTANE included 7,669 adults who started oral semaglutide between January 5 and February 7, 2026, and remained on treatment for three months. Unlike SIGNAL, it allowed previous weight-management treatment. Paired baseline and three-month food-noise information came from 6,180 people.
The reported average score fell from 2.8 to 2.1, a change the sponsor described as approximately 25%. This is a relative change in a questionnaire score. It is not a finding that people had 25% fewer thoughts about food, nor that one quarter of participants stopped experiencing intrusive thoughts.
Questionnaire averages summarize responses on a rating scale, rather than counting thoughts. Because the ratings start at 1, the percentage also depends on the scale’s starting point. The more informative description keeps the actual scores, three-month interval and selected population together. Individual experiences may differ from the average.
The missing voices are part of the result
Neither study is described as having a randomized untreated control group. Changes could reflect medication, accompanying care, expectations or other changes over time. Statistical adjustment can address measured factors; it cannot recreate random assignment. Our guide to randomized and real-world evidence explains this difference.
Requiring treatment persistence also filters out experiences. People who discontinue because of side effects, cost or limited benefit may differ from those who continue. Routine telehealth questionnaires can be incomplete or collected at uneven times. The sponsor acknowledges selection bias, self-report limitations and restricted generalizability. Its summary does not establish the experience of people with diabetes or eating disorders, who were excluded.
Earlier research illustrates another measurement problem. The INFORM survey, published in May 2026, asked 550 current injectable-semaglutide users about present food noise and their recollection of it before treatment. It used a different five-item questionnaire. Its authors acknowledge recall and selection bias; most respondents were white women. Those results provide context, not a directly comparable score or independent replication of SIGNAL and OCTANE.
Useful for a conversation, insufficient for a treatment ranking
The pill and injection findings cannot show which formulation better reduces food noise. Follow-up periods, treatment histories and analyses differ, and participants were not assigned head-to-head. The semaglutide profile provides broader medicine context; this announcement should not become a reason to change treatment independently.
Food-related distress can still be worth discussing. Concrete examples—difficulty concentrating at work, interrupted social meals or persistent worry about eating—can help describe the problem without turning a questionnaire average into a personal target. The appointment guide can help organize questions about daily functioning alongside weight, nutrition and tolerability.
The current US prescribing information includes defined weight-management indications, not a separate indication for food noise. It also describes common gastrointestinal adverse effects and a boxed thyroid C-cell tumor warning based on rodent findings, with uncertain human relevance. A perceived benefit needs to be considered alongside these risks and contraindications.
The next useful evidence would include fuller conference reports, follow-up counts, outcomes after discontinuation and comparisons that can better separate treatment effects from selection. This explainer reviewed the complete sponsor announcement, earlier measurement and survey papers, and relevant label sections. The SIGNAL and OCTANE posters and full manuscripts were not inspected, so claims requiring their detailed tables remain unresolved.



