The short answer
A weight-reduction percentage is uninterpretable on its own. The same trial can legitimately report two different figures depending on the estimand, and figures from different trials differ by population, duration, comparator and dropout rate before they differ by compound. Five checks make the numbers readable.
Key facts
- Check 1
- Which estimand (efficacy or treatment-policy)
- Check 2
- Population (comorbidities change the result)
- Check 3
- Duration (curves may not have plateaued)
- Check 4
- Comparator (placebo or active)
- Check 5
- Discontinuation rate, not just efficacy
- Worked example
- CagriSema 22.7% vs 20.4%, same trial
Estimands: the reason one trial reports two numbers
REDEFINE 1 reported both 22.7% and 20.4% mean reduction for CagriSema. Neither is wrong. The first is the efficacy estimand, describing participants who adhered to treatment. The second is the treatment-policy estimand, describing everyone randomised regardless of whether they continued. Efficacy answers 'how well does this work when taken'; treatment-policy answers 'what happens if we prescribe it'. Press coverage almost always quotes the larger figure without naming which it is.
Population changes the answer more than potency does
Retatrutide's TRIUMPH-1 reported 28.3% in a relatively uncomplicated obesity population. TRIUMPH-3, in adults with severe obesity and established cardiovascular disease, reported up to 22.6%. Same compound, same sponsor, similar duration. The gap is population, not potency. Comparing a figure from a healthy-cohort trial against one from a comorbid cohort tells you almost nothing about the compounds.
Research material referenced
Retatrutide 10mg, third-party HPLC tested
Duration, and whether the curve had flattened
A percentage is meaningless without its timepoint, and a timepoint is more informative when you know whether reduction had plateaued. Amycretin's Phase 2 reported that at higher doses no plateau had been reached by week 36, which means the figure reflects the trial's length. A trial that stops while the curve is still descending reports a floor for that duration; one that ran past plateau reports something closer to a ceiling.
Discontinuation belongs next to efficacy
TRIUMPH-1 reported 11.3% discontinuation due to adverse events on the 12 mg arm against 4.9% on placebo. That figure is routinely dropped from summaries, and it is exactly the number that connects to the estimand question. The people who discontinue are the difference between the two headline percentages. A compound with a larger effect and a much higher dropout rate is not straightforwardly better.
Placebo response is not zero
Placebo arms in these trials typically show 2-3% reduction, because trial participation itself involves dietary and activity counselling. Some summaries quote absolute reduction and others quote placebo-adjusted reduction, and the difference is a couple of percentage points. Amycretin's Phase 2, for instance, was reported as 11.9% placebo-adjusted for the subcutaneous arm rather than as an absolute figure.
A checklist
Before comparing any two figures, confirm they match on the following. If they do not, the comparison is indirect at best.
- Same estimand (efficacy or treatment-policy)
- Comparable population and baseline BMI
- Same or similar duration, and whether the curve plateaued
- Same comparator, and whether the figure is placebo-adjusted
- Discontinuation rates reported alongside efficacy
Frequently asked questions
What is an estimand?
A precise statement of what a trial is measuring, including how it handles participants who stop treatment. It determines which participants contribute to the headline figure.
Why do placebo groups lose weight?
Trial participation itself includes dietary and activity support, so the placebo arm typically shows 2-3% reduction.
Can I compare retatrutide's 28.3% with CagriSema's 22.7%?
Only indirectly. The durations differ, the populations differ and the estimands may differ. The comparison is suggestive, not precise.
Continue researching
Related guides and reference material.
Primary sources & clinical trials
Published papers, trial registrations, regulator pages and company announcements. Links open the original.
Written and published by UK Peptides, the trading name of AURYXRESEARCH LTD (company 17388727, registered in England and Wales).
Questions or a correction? Email hello@ukpeptides.org or call +44 7426 987848.
More GLP-1 & Incretin Science articles
- Oral Semaglutide vs OrforglipronBoth are tablets, only one is a peptide. Why oral semaglutide needs SNAC and a fasting window while orforglipron needs neither, and what that costs each.
- What Is Aleniglipron? The Second Oral Small-Molecule GLP-1Aleniglipron is an oral non-peptide GLP-1 agonist from Structure Therapeutics. ACCESS reported 12.1% at 36 weeks; ACCESS II reported 15.3% absolute at 44 weeks.
- What Is Semaglutide? Structure, Mechanism and EvidenceSemaglutide is a GLP-1 analogue with 94% homology to native GLP-1. Its structure, the C18 diacid that gives it a weekly half-life, and the SELECT outcome data.
- What Is Tirzepatide? The First Dual Incretin AgonistTirzepatide is a dual GIP and GLP-1 receptor agonist built on a GIP backbone. SURMOUNT-1 reported up to 22.5% weight reduction over 72 weeks.
- Semaglutide vs Tirzepatide: What the Direct Comparison ShowedSURMOUNT-5 randomised 751 adults to tirzepatide or semaglutide for 72 weeks: 20.2% against 13.7% mean reduction. The full comparison, including tolerability.