Faculty guide: reading the weight-management trial comparison
Version 1.1.0 (2026-11-14) · Flagship Weight Management 1.4.0 · AI-assisted source verification; human review none recorded. Not yet run. Human review and the pilot are pending until real participants complete the tasks and a named person records the results.
Goal. Students learn to tell a head-to-head trial from a cross-trial comparison, to say what a result supports and what it does not, and to reopen the exact result from a copied citation.
Time. About 65 minutes (50 minutes if items 8 and 9 are set as homework). Needs a browser and a printed or on-screen worksheet.
Before class
- Open the worksheet and the instructor key. The key is a public, unlisted page: share it with instructors, not in the student link.
- Confirm the release: every source link names Flagship Weight Management 1.4.0. If a newer release has shipped, these links still return the version students were given.
- Skim the five results on the comparison page so you know which one is the only head-to-head trial (SURMOUNT-5).
Class flow
- 5 min. Frame the task: a classmate quotes a bigger percentage from a different trial. Can it be compared?
- 20 min. Students work items 1 to 3 in pairs using the comparison page and the Verify this result panel. Debrief: design first, numbers second.
- 20 min. Items 4, 5 and 8: estimands, intervals and treatment discontinuation. Use the key's wording for the debrief.
- 15 min. Items 6 and 7: the observational cohort, then copy a citation and reopen it in a private window; open the 1.0.0 address and find the pointer to the update.
- 10 min. Item 9: write a supported statement, give the claim a verdict, then open the claim check and compare.
Common misreadings to listen for
- Treating a bigger percentage in one trial as proof against a different trial (the central error of the lesson).
- Reading a confidence interval as the range in which most patients fall.
- Reading study completion as treatment adherence.
- Assuming a registry result and the paper report the same analysis (SURMOUNT-1 shows they can differ).
These are anticipated before any pilot. They are not observations. The pilot sheet records what actually happens.
Where each comparison row was checked (release 1.4.0)
| Result | Checked against | Matched / disclosed differences | Not read |
|---|---|---|---|
| SURMOUNT-5 (phase 3b, open-label) | ClinicalTrials.gov results; PubMed abstract; Statistical analysis plan | 4 / 0 | Publisher page |
| STEP 1 (double-blind) | PubMed abstract; ClinicalTrials.gov results; Statistical analysis plan | 3 / 1 | Publisher page |
| SURMOUNT-1 (phase 3, double-blind) | PubMed abstract; ClinicalTrials.gov results; Statistical analysis plan; Study protocol | 3 / 1 | Publisher page |
| Retatrutide phase 2 (double-blind) | PubMed abstract; ClinicalTrials.gov results; Statistical analysis plan | 5 / 0 | Publisher page |
| Rodriguez and colleagues, JAMA Internal Medicine 2024: cohort study, not a randomized trial | Open-access full text | 9 / 0 | Publisher page |
- The four journal papers behind the trial rows could not be opened by WPF's automated access (HTTP 403), so the exact table or figure for those numbers is not located; students check the registry results page and the analysis plan instead. The observational cohort was checked against its open-access full text.
- The lesson teaches how to read evidence. It is not medical advice and recommends no treatment.