Instructor key

Version 1.1.0 (2026-11-14), against Flagship Weight Management 1.4.0. Back to the student worksheet.

  • Time: about 65 minutes. Items 1 to 3 take 20 minutes, items 4, 5 and 8 take 20, items 6 and 7 take 15, item 9 takes 10. For a 50-minute class, assign items 8 and 9 as homework.
  • Students need only a browser. Every link is a versioned address, so the exercise stays the same after newer releases are published.
  • The key is a public page and is not hidden. It is not linked from the worksheet; share it with instructors directly.
  • These materials teach reading evidence. They are not medical advice and do not recommend any treatment.
  • AI-assisted source verification; human review none recorded. If your students find an error, report it through the corrections page.
  1. Question 1

    Open the five results. For each, write whether the two drugs were compared with each other inside one randomized trial. Which result is the only one that does so?

    Only SURMOUNT-5 (surmount-5-weight-72w) is a direct head-to-head randomized comparison of tirzepatide and semaglutide. STEP 1 and SURMOUNT-1 each compare one drug with placebo; the retatrutide phase 2 trial is also placebo-controlled; the JAMA Internal Medicine study is an observational cohort in which patients were not randomized.

    Sources: SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.4.0 · STEP 1 (double-blind), record step-1-weight-68w, release 1.4.0 · SURMOUNT-1 (phase 3, double-blind), record surmount-1-weight-72w, release 1.4.0 · Retatrutide phase 2 (double-blind), record retatrutide-phase2-weight-48w, release 1.4.0 · Rodriguez and colleagues, JAMA Internal Medicine 2024: cohort study, not a randomized trial, record jama-intern-med-2024-realworld, release 1.4.0

  2. Question 2

    A classmate says: "Tirzepatide beat semaglutide by 6.5 percentage points." Find the numbers in SURMOUNT-5. Where does 6.5 come from, and which design makes the subtraction allowed?

    -20.2% with tirzepatide vs -13.7% with semaglutide (least-squares means; P<0.001). The difference is 20.2 − 13.7 = 6.5 percentage points. The subtraction is allowed because both drugs were randomized in the same trial (72 weeks, 751 adults with obesity, without type 2 diabetes), so the groups are comparable by design. The 95% intervals (95% CI -21.4 to -19.1 (tirzepatide); -14.9 to -12.6 (semaglutide)) do not overlap. The trial was open-label, which is a limit.

    Sources: SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.4.0

  3. Question 3

    The same classmate then subtracts the STEP 1 semaglutide result from the SURMOUNT-1 tirzepatide result and says it "proves the same thing". Write down the population, follow-up and comparator of each. Is the subtraction valid? Why or why not?

    Not valid as a head-to-head. STEP 1: 1,961 adults with BMI 30 or more (27 or more with a weight-related condition), without diabetes, 68 weeks, Placebo plus lifestyle intervention (2:1 randomization); result -14.9% with semaglutide vs -2.4% with placebo; estimated difference -12.4 percentage points. SURMOUNT-1: 2,539 adults with BMI 30 or more, or 27 or more with a weight-related complication, excluding diabetes, 72 weeks, Placebo (1:1:1:1 randomization), including a 20-week dose-escalation period; result -15.0%, -19.5% and -20.9% with 5, 10 and 15 mg vs -3.1% with placebo. Different populations, durations and designs make this an indirect comparison across trials. WPF's own explanation: "This drug was compared with placebo in its own trial. Setting two trials side by side is an indirect comparison across different populations, durations and designs, so their percentages cannot be subtracted."

    Sources: STEP 1 (double-blind), record step-1-weight-68w, release 1.4.0 · SURMOUNT-1 (phase 3, double-blind), record surmount-1-weight-72w, release 1.4.0

  4. Question 4

    Read the estimand for SURMOUNT-1. In your own words, what question does the displayed number answer? Then open the 1.0.0 version of the same result and write down what it said about the analysis plan.

    Release 1.4.0 says: The abstract states a treatment-regimen estimand (regardless of treatment discontinuation, intention-to-treat). The analysis plan calls the matching estimand "hybrid" (regardless of adherence; ANCOVA with multiple imputation, §6.13.1.2). The registry's posted results (-16.0%, -21.4%, -22.5% vs -2.4%) are a different, efficacy-style analysis that excludes data after drug was stopped, so they do not check the displayed values. The plan also says publications would follow the efficacy estimand, and the protocol assigns the efficacy estimand to all purposes other than the FDA submission, which differs from the abstract; this is reported, not resolved. Release 1.0.0 labelled it a treatment-regimen estimand and stated the public analysis plan had not been reviewed. The displayed percentages did not change between versions; what changed is the disclosure of which analysis produced them and that the registry's posted results are a different analysis.

    Sources: SURMOUNT-1 (phase 3, double-blind), record surmount-1-weight-72w, release 1.4.0 · SURMOUNT-1 (phase 3, double-blind), record surmount-1-weight-72w, release 1.0.0

  5. Question 5

    SURMOUNT-5 reports tirzepatide −20.2% with a 95% interval of −21.4 to −19.1. What does the interval tell you, and what does it not tell you?

    It describes the precision of the estimated average change for the trial population: the data are compatible with true average changes in that range. It does not say that every person loses that much, that results generalize to people outside the population (751 adults with obesity, without type 2 diabetes), or anything about bias from the open-label design.

    Sources: SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.4.0

  6. Question 6

    The JAMA Internal Medicine cohort found tirzepatide users more likely to reach 5%, 10% and 15% weight loss. Name two reasons this is weaker evidence for "tirzepatide works better" than SURMOUNT-5, then name one thing it can still add.

    Patients were not randomized. People who chose or were prescribed each drug were matched statistically, so differences can reflect factors that were not measured. Results describe time on treatment only. Reasons: patients were not randomized, so unmeasured differences can explain the gap; results describe time on treatment only, so people who stopped are not counted. It can still add information about outcomes in everyday practice in a much larger group (41,222 US adults with overweight or obesity on semaglutide (32,029) or tirzepatide (9,193); 18,386 after propensity-score matching; 52.0% of the matched group had type 2 diabetes), and its direction can be compared with the trial result.

    Sources: Rodriguez and colleagues, JAMA Internal Medicine 2024: cohort study, not a randomized trial, record jama-intern-med-2024-realworld, release 1.4.0

  7. Question 7

    Copy the citation for SURMOUNT-5 from its record page. Open the versioned-record address in a private browser window. Write down the release version and the resultsSha256 shown, then open the same address with version=1.0.0 and note what the response says about the current version.

    The 1.4.0 address returns release wpf-flagship-weight-management, version 1.4.0, result surmount-5-weight-72w and a 64-character resultsSha256. The 1.0.0 address returns archived: true, names the current version and points to updatedRecordUrl and the What changed? page. A citation that names a version keeps resolving to that version.

    Sources: SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.4.0 · SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.0.0

  8. Question 8

    How did SURMOUNT-5 and the retatrutide phase 2 trial handle people who stopped taking the study drug? Write one sentence for each. Then say what the registry participant flow does and does not tell you about stopping the drug.

    SURMOUNT-5: Handled by the primary estimand: participants stay in the analysis regardless of stopping the study drug or taking other anti-obesity medication, except switching to the other study drug. Registry participant flow counts study completion, not stopping the drug: 319 of 375 (tirzepatide) and 319 of 376 (semaglutide) completed the study. The registry posts no count of participants who stopped the drug and stayed in the study. Retatrutide phase 2: Handled by a hypothetical strategy: the primary (efficacy) estimand asks what the effect would be had participants adhered, and the registry's 48-week results use a model that excludes data after permanent discontinuation. Registry participant flow counts study completion, not stopping the drug: placebo 50 of 70, 1 mg 60 of 69, 4 mg 52 of 67, 8 mg 59 of 70, 12 mg 54 of 62 (by starting-dose arm). The registry posts no count of participants who stopped the drug and stayed in the study. The participant flow counts people who completed or left the STUDY. Someone can stop the drug and stay in the study, so study completion is not the same as treatment discontinuation, and neither registry posts the second count.

    Sources: SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.4.0 · Retatrutide phase 2 (double-blind), record retatrutide-phase2-weight-48w, release 1.4.0

  9. Question 9

    A classmate writes: "Tirzepatide caused more weight loss than semaglutide, because its trial showed a bigger percentage." (a) Write two sentences about SURMOUNT-5 that are supported by the record and say nothing more. (b) Which verdict from Peptide Claims & Evidence fits the classmate's sentence: Supported within scope, Overstated, Insufficient evidence or Contradicted by evidence? Explain in one sentence, then open the claim check and compare.

    (a) A model answer: In SURMOUNT-5, a 72 weeks open-label randomized trial of 751 adults with obesity, without type 2 diabetes, -20.2% with tirzepatide vs -13.7% with semaglutide (least-squares means; P<0.001). Because both drugs were randomized in the same trial, that difference is a head-to-head result for that population, with the open-label design as a limit. (b) Overstated: the numbers support a head-to-head difference in this one trial, but "caused", "because its trial showed a bigger percentage" and any wider population go beyond it, and percentages from separate trials cannot be compared this way. The claim check for "when weight-loss results can be compared" gives that verdict.

    Sources: SURMOUNT-5 (phase 3b, open-label), record surmount-5-weight-72w, release 1.4.0