Why one GLP-1 study can report two weight-loss results

The number answers a question defined by the study

Researchers use the term “estimand” for the treatment effect they want to estimate. FDA’s statistical guidance asks researchers to define that question clearly, including how they handle events such as stopping treatment or taking another medicine. The analysis should match the question.

Imagine a fictional participant who stops a trial medicine midway through the study but returns for the final weight measurement. The measurement exists. Researchers still need a plan for whether it helps answer the question they are studying. Stopping treatment and disappearing from follow-up are different situations.

Before comparing two percentages, ask what each number is intended to describe. The name of the medicine and the week of follow-up do not tell you the whole analysis plan.

Two common ways of describing GLP-1 trial effects

A treatment-policy question can ask about the effect of assignment to treatment regardless of whether participants later stop it or use another weight-loss therapy. It keeps those events within the question being answered.

A trial-product question may instead ask about an idealized scenario in which participants continue their assigned treatment and do not use other weight-loss therapies. That is a hypothetical effect estimated under stated assumptions; it is not simply a promise about everyone who keeps taking a medicine.

These descriptions match the two analyses in the manufacturer’s OASIS 4 summary. Terminology and details vary between studies. Read the explanation attached to the figure instead of assuming that every “on-treatment” result uses the same method.

A real example: the OASIS 4 oral semaglutide trial

The OASIS 4 registry record identifies a randomized phase 3 study involving 307 participants, with weight outcomes assessed at week 64. It studied oral semaglutide against placebo. The current Wegovy label describes the adult population without diabetes and the diet and activity support supplied during the trial.

The manufacturer reports an estimated 16.6% mean weight reduction under the scenario in which participants stayed on treatment without other weight-loss therapies. The current label reports a 13.6% reduction for the tablet group in its intention-to-treat analysis. Those figures concern the same trial but do not describe the same treatment-effect question.

The three-percentage-point gap is not an estimate of what one missed tablet costs an individual. It does not tell someone how much weight they would lose with perfect adherence. Keep the assumptions attached to the averages, and compare each with the corresponding placebo result from the same analysis.

Our Wegovy pill trial overview connects these results to practical care questions. Comparing an oral-trial average with an injection-trial average also requires attention to the different participants, follow-up and analysis; those numbers alone cannot establish clinical equivalence.

“All randomized participants” does not mean every measurement was observed

In its tablet trial table, the Wegovy label states that some week-64 weight measurements were missing and describes statistical imputation. That means the analysis used a method to estimate missing information under assumptions, rather than having a measured final weight for every participant.

FDA’s guidance distinguishes the handling of treatment changes from the handling of missing data. A participant can stop medication and still provide measurements; someone else can remain assigned to treatment but miss a visit. Sensitivity analyses examine how conclusions depend on assumptions.

When a summary says “everyone was included,” look for what the full report says about missing measurements and follow-up. The phrase should not erase the uncertainty involved. You do not need to perform the calculation yourself to notice when an explanation is absent.

Five details to carry with a study result

Write a short description of the result before putting it beside another treatment headline. Our general GLP-1 headline checklist helps establish the event date and source; these questions add the trial-analysis detail.

Who participated, and which people were excluded?
Which formulation was studied, and what was the comparison group?
At what follow-up point was the weight change assessed?
How did the analysis handle stopping treatment and missing measurements?
What do the safety findings and discontinuations add to the weight average?

Use the evidence to ask about care, not to select a promised percentage

CoreAge Rx describes compounded semaglutide and tirzepatide on its official site. OASIS 4 studied an oral semaglutide trial product; it did not study CoreAge Rx patients or establish outcomes for its compounded prescriptions. Ask which evidence the clinician considers relevant to the exact treatment proposed.

The service’s intake process asks for health history and medicines. Its refill process includes check-ins and provider review. These are useful places to ask how benefit, side effects and problems continuing treatment would be discussed. A trial average cannot tell you whether a service will meet those needs.

Use our CoreAge Rx review to investigate the program with those questions in hand. Agree on goals and follow-up with a clinician instead of choosing a provider because an advertisement repeats the largest study figure.

FDA’s compounding guidance remains relevant: compounded GLP-1 medicines are not FDA-approved and should be used only when an approved drug cannot meet a patient’s medical needs. Clinical-trial headlines do not remove that separate product question.

Common questions

Is the larger trial estimate automatically the better one to use?

Use the estimate that answers the question you are asking, with its assumptions and limitations stated. A larger number is not a reason to leave out the other analysis.

Is a treatment-policy result the same as a real-world study?

No. It can account for treatment changes within a randomized trial, but the trial still has its own eligibility rules, visits and support. That does not turn it into a study of ordinary care.

Does the study average set my personal weight-loss target?

No. An average describes a group under the study conditions. Discuss a suitable goal, tolerability and follow-up with your clinician.

Sources and fact-checking

Sources checked 2026-09-09. CoreAge Rx pages support descriptions of its service; medical and regulatory sources support the educational context. Offers and availability can change.