Two weight-loss percentages can sit next to each other in a headline even when the participants were never in the same trial. That is the first thing to check in the latest tirzepatide–semaglutide comparison.
What the September announcement reported
On 29 September 2026, Lilly described an indirect comparison of tirzepatide in SURMOUNT-1 and semaglutide 7.2 mg in STEP UP. The model adjusted for age, sex, starting weight and baseline HbA1c.
| 72-week comparison against semaglutide 7.2 mg | Additional weight reduction, treatment-regimen estimand | 95% credible interval for the difference |
|---|---|---|
| Tirzepatide 15 mg | 4.5 percentage points | −6.3 to −2.8 points |
| Tirzepatide 10 mg | 3.2 percentage points | −5.0 to −1.4 points |
Negative differences favor tirzepatide because the outcome is change in body weight. These figures use the treatment-regimen estimand, which includes effects regardless of treatment discontinuation.
Under the efficacy estimand, the differences were −1.8 points (95% credible interval −3.8 to 0.1) and −0.7 points (−2.6 to 1.3). Neither was statistically significant. The conclusion therefore depends on the analysis being quoted.
An indirect comparison, in ordinary language
Picture two schools testing different teaching methods. Comparing their exam results may be useful, but the schools could also differ in their pupils, attendance or testing conditions. Statistical adjustment tries to account for some differences; it cannot guarantee that every relevant difference has been removed.
A head-to-head randomized trial assigns people within one study to the treatments being compared. An indirect comparison instead links information from separate studies. These approaches answer related questions with different levels of control over the comparison.
Why two analyses can give different answers
An estimand is simply the precise treatment-effect question a study sets out to answer. One question may include what happens after people stop treatment. Another may focus on the effect before discontinuation. A different question can produce a different estimate.
For this reason, quoting only the larger difference would give an incomplete account. A readable explanation should show both sets of findings without burying the less favorable result.
What does the interval tell us?
The table uses credible intervals, not confidence intervals. For the efficacy estimates, the reported ranges cross zero. Zero represents no difference, so those estimates do not provide a clear statistical separation under that analysis.
An interval is not a promise about an individual’s result. It describes uncertainty around an estimated group difference within the model.
A checklist for the next comparison headline
- Were the treatments tested together or in different trials?
- Were the populations and follow-up periods comparable?
- Which estimand produced the number?
- Does the uncertainty interval include no difference?
- Are adverse events and discontinuations reported alongside weight change?
Those five questions are more useful than treating a single percentage as a ranking of every product with the same molecule name.
Related materials for laboratory research
The NHD Tirzepatide page is the starting point for identifying laboratory catalog options. The analysis discussed above concerns clinical trial medicines, not tests of NHD material.
For an enquiry, tell us the material name, quantity and planned laboratory test. Ask which identity, purity and other analytical records are available for the selected variant. Request product information from NHD.
The medicines used in these trials are not interchangeable with NHD catalog materials. Trial findings do not establish clinical effects, formulation equivalence or the quality of an NHD batch.
Read more
Explore the related research explanation.
Source and editorial review
Lilly announcement of EASD analysis, 29 September 2026. Accessed 9 October 2026. This article explains a company announcement; it is not independent peer review. AI-assisted source and readability checks were completed before publication.
NHD catalog materials are for laboratory research, development and manufacturing, not human or veterinary use. This article provides no treatment or dosing instructions.
Sources & editorial method
AI-assisted writing
0 linked records are listed in the references below. Read the editorial and AI-assistance policy.
How to interpret this article
AI-assisted source and readability checks, not independent peer review.
Research-use boundary: Catalog materials discussed on this website are for laboratory research, development and manufacturing use only, not for human or veterinary use. This content is not medical advice and does not provide administration instructions.
Continue with structured records
Move from this editorial analysis to the product identity, filtered evidence and controlled terminology behind it.
