Percent change starts from baseline weight
In a weight-loss study, percent change expresses the difference relative to a participant’s starting weight. A group mean summarizes that outcome across the study’s analysis, rather than promising that each person will reach the mean.
SURMOUNT-5 reported mean changes of −20.2% and −13.7% at week 72 for its studied regimens. The population, dose and analysis belong beside those numbers.[source]
Percentage points are not relative percent improvement
Subtracting 13.7 from 20.2 gives a 6.5-percentage-point difference. Expressing that difference relative to 13.7 produces approximately 47% greater relative weight reduction. These describe the same comparison using different denominators.
The relative figure does not mean participants lost 47% of their starting weight. Our chart uses the direct group values to keep that distinction visible.[source]
A threshold result asks a different question
A responder rate reports the proportion of participants reaching a specified threshold. The sponsor reports that 64.6% of the tirzepatide group and 40.1% of the semaglutide group achieved at least 15% weight loss.[source]
Those numbers are percentages of participants, not average percentages of body weight lost. Switching chart views changes the question being displayed, so the label and axis must change too.
Confidence intervals describe uncertainty around an estimate
A confidence interval is not the minimum and maximum response experienced by patients. It reflects uncertainty in the group estimate under the analysis method. Individual responses can be much more variable.
Analysis choices also matter when treatment is stopped or data are missing. The sponsor identifies the main SURMOUNT-5 figures here as treatment-regimen estimand results; they should not be mixed casually with a different trial’s analysis.[source]
Numbers do not travel without their study context
A result from one dose and formulation cannot automatically describe another formulation, a compounded product or a telehealth provider. Unrelated trials can differ in populations, follow-up and methods.
Read our full trial explanation for the boundaries. The chart is a guide to interpreting evidence, not a personal outcome calculator or a prescribing recommendation.
Sources & further reading
Provider pages establish what a brand markets. Clinical and regulatory sources establish the limits of the evidence. Each source has its own last-checked date below.
- Aronne et al., NEJM 2025: tirzepatide compared with semaglutide for obesityPrimary randomized trial abstract; SURMOUNT-5, 751 adults without diabetes, 72 weeks; funded by Eli Lilly · Checked October 8, 2026
- Lilly: complete SURMOUNT-5 results and trial designTrial sponsor report; treatment-regimen estimand and safety-comparison limitations · Checked October 8, 2026