SURMOUNT-5 directly compared tirzepatide with semaglutide for weight management in adults with obesity without type 2 diabetes. That randomized comparison is more informative for its specific question than placing percentages from unrelated trials side by side. It still has boundaries: an open-label design, assigned maximum-tolerated regimens, a defined follow-up and a weight endpoint. The result cannot choose a medicine for every individual or establish a head-to-head cardiovascular outcome.
Start with the intervention and the population
Semaglutide is a GLP-1 analogue, while tirzepatide has GIP and GLP-1 receptor activity. The SURMOUNT-5 publication randomized 751 adults with obesity without diabetes to the specified maximum-tolerated treatment strategies. The primary endpoint was percentage change in body weight at week 72.
The absence of diabetes in this population is part of the result. It should not be silently generalized to every diabetes setting. Likewise, the comparison describes the studied formulations and strategies, not every product carrying either molecule's name.
What the weight findings mean numerically
The published mean weight changes were −20.2% with tirzepatide and −13.7% with semaglutide. The difference between those means is 6.5 percentage points. Both groups were measured within the same randomized experiment, which makes the contrast more informative than comparing unrelated trial headlines.
Scroll horizontally to read all columns.
| Expression | Meaning |
|---|---|
| −20.2% and −13.7% | Reported mean changes within the two groups |
| 6.5 percentage points | Difference between the reported means |
| 72 weeks | Assessment horizon |
| Group average | A study estimate, not an individual prediction |
Calling this a 6.5% difference without explaining the denominator can confuse a percentage-point contrast with a relative percentage.
What randomization adds, and what open-label means
Random assignment helps make the groups comparable at the start. SURMOUNT-5 was open-label, so the assigned intervention was not masked in the same way as in a double-blind study. Knowledge of treatment can influence expectations and behavior, even when body weight is measured objectively.
That does not erase the randomized result. It is a design feature to preserve when describing the strength and limitations of the evidence. Our critical-reading guide explains why randomization and blinding should be described separately rather than treated as one quality label.
Averages and responder thresholds answer different questions
An average weight change summarizes the distribution but does not reveal every participant's outcome. Threshold results, such as the proportion reaching a specified percentage reduction, add another perspective. They still do not tell an individual what will happen to them.
Read the handling of missing observations and treatment discontinuation before comparing either measure. A trial can report more than one estimand, each representing a different question about adherence, rescue treatment or discontinuation. Preserve the analysis definition that belongs to the number being cited.
Safety and tolerability remain part of the comparison
Gastrointestinal adverse events were the most common events in both groups in the publication. A reader should also examine severity and discontinuation rather than treating common events as the whole safety story. Rare events and long-term effects may need evidence beyond a single trial.
The current WEGOVY prescribing information is a separate source for formulation-specific indications, warnings and pharmacology. A head-to-head efficacy result does not replace either product's full safety record. This article does not recommend starting, stopping or switching medicines.
A worked example of percentage-point confusion
Imagine a summary says that treatment A produced 20% mean reduction and treatment B produced 14%. The absolute difference is six percentage points. Dividing six by fourteen describes a relative comparison of those reductions, which is a different numerical statement. Neither calculation tells the reader how many additional people reached a responder threshold.
This example explains why an article should state the original group values and the unit of comparison. It avoids a dramatic relative framing that can sound like the participant lost that percentage of their entire starting weight beyond the comparator.
Why weight and clinical events need separate comparisons
Suppose one intervention has a larger mean weight reduction in a direct comparison, while another has an event-based trial in a different population. These results should not be merged into one overall superiority statement. The outcomes, populations and study questions differ.
A useful comparison states what SURMOUNT-5 directly measured and then identifies other clinical questions requiring separate evidence. This prevents weight change from becoming a stand-in for every cardiovascular, renal or quality-of-life outcome. It also avoids the opposite mistake of treating a well-defined weight result as uninformative because it does not answer every question.
The strength of a study is its ability to support the conclusion it was designed to test, with its limitations made clear.
The comparison is between assigned strategies, not isolated receptor counts
The trial assigned each medicine under its specified maximum-tolerated regimen. The intervention includes how participants were escalated and what exposure they could tolerate. It is not an experiment holding every pharmacological variable constant while changing only the number of receptor targets. The causal contrast concerns the assigned treatment strategies.
This distinction helps explain why a dual-receptor description is not a complete explanation of the observed difference. The trial establishes the clinical comparison it randomized; isolating the contribution of each receptor activity would require another design or appropriately qualified mechanistic analysis.
How to read the weight result beyond one mean
The mean percentage change summarizes the group. Responder thresholds describe proportions reaching specified amounts of weight reduction. Waist circumference is another measurement. Read these together to understand the distribution without pretending that every participant experienced the mean.
Also identify the estimand and handling of discontinuation. A result intended to describe assignment under a treatment-policy strategy can differ from one describing continued adherence. The analysis label should remain beside the figure, particularly when a secondary source quotes only the larger estimate.
Open-label design and tolerability need a balanced reading
Open-label means participants and investigators knew the assigned medicine. Body weight is an objective measurement, but knowledge of assignment can still matter for behavior, reporting and treatment experience. The design should be acknowledged without dismissing the randomized weight comparison.
Adverse events and discontinuation describe treatment burden in addition to efficacy. A larger mean reduction is not a complete measure of individual suitability. Existing conditions, contraindications and other clinical factors belong to a product-specific assessment that this article cannot reconstruct.
The study’s weight endpoint also does not provide a direct comparison of every disease outcome. SELECT, FLOW and SUMMIT address other questions in different populations. Their results should not be combined into a synthetic head-to-head heart or kidney ranking. For readers of research peptides, SURMOUNT-5 is a strong example of a direct comparison whose value is greatest when its actual question remains intact.
What this trial cannot settle
SURMOUNT-5 does not compare every oral or injectable formulation, every possible population or every clinically important outcome. In particular, a weight endpoint is not automatically a direct comparison of cardiovascular events, kidney failure or long-term quality of life.
For a different evidence problem, read retatrutide versus tirzepatide. That comparison requires extra care because newer placebo-controlled retatrutide publications are not themselves head-to-head trials against tirzepatide. Matching the question to the study is more valuable than ranking receptor counts.
