# What GLP-1 Weight-Loss Trial Results Actually Mean

How to read the headline numbers from the semaglutide and tirzepatide weight-loss trials: averages, placebo, who was enrolled, how long, and why a trial average is not a forecast for you.

Canonical HTML: [What GLP-1 Weight-Loss Trial Results Actually Mean](https://joinelovi.com/blog/what-glp-1-weight-loss-trial-results-actually-mean)

Prepared by [Ēlovi Editorial](https://joinelovi.com/blog/authors/elovi-editorial).
Published: 2026-10-11T14:40:00.000Z. Updated: 2026-10-11T14:40:00.000Z.
Prepared by Ēlovi Editorial with AI assistance. Sources checked: 2026-10-11. Reviewed by the Ēlovi editorial team, which includes licensed clinicians.
[Editorial process](https://joinelovi.com/blog/editorial-policy).

## The short answer

A trial result such as 'average weight loss of 14.9%' is the group's average change at a fixed time point, in people who met the trial's entry rules and received diet and activity counseling, measured against a placebo group that lost 2.4%. It is not a forecast for one person, and it is not the result of a compounded product. In the 72-week tirzepatide trial, 85% of people on the lowest studied dose lost at least 5% but 30% lost at least 20%, so individual results spread widely around the average.

## What does "average weight loss" in a GLP-1 trial actually mean?

It means the average change in body weight across everyone randomized to a group, measured at one fixed time point. In the semaglutide trial known as STEP 1, that was a mean change of −14.9% at week 68, compared with −2.4% for the placebo group. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) In the tirzepatide trial known as SURMOUNT-1, the three studied dose groups averaged −15.0%, −19.5%, and −20.9% at week 72, compared with −3.1% for placebo. [2](https://pubmed.ncbi.nlm.nih.gov/35658024/)

Two things follow. The average describes a group, not a person: some people lost more and some lost less. And the number is a snapshot at 68 or 72 weeks, after a dose-escalation period of up to 20 weeks in the tirzepatide trial. [2](https://pubmed.ncbi.nlm.nih.gov/35658024/) It says nothing about month three, and nothing about what happens after treatment ends.

This page explains how to read those numbers. It cannot tell you what your own result will be. If you are comparing the medicines themselves, start with our [guide to GLP-1 treatment](https://joinelovi.com/blog/glp-1-guide) and the comparison of [semaglutide and tirzepatide](https://joinelovi.com/blog/semaglutide-vs-tirzepatide).

## Why does the placebo group matter?

The placebo group shows what happened to people who got the same trial attention, the same counseling, and an inactive injection. In both trials, everyone received lifestyle support: instruction on a reduced-calorie diet (about a 500 kcal/day deficit) and counseling to reach at least 150 minutes of activity a week. [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b) [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b) Placebo participants still lost weight, an average of 2.4% in STEP 1 and 3.1% in SURMOUNT-1. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) [2](https://pubmed.ncbi.nlm.nih.gov/35658024/)

So the effect attributed to the medicine is the gap, not the headline. In STEP 1 the estimated difference was 12.4 percentage points. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) In SURMOUNT-1 it ranged from 11.9 to 17.8 points across the three dose groups. [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b) Quoting only the treated group's average overstates what the medicine alone did and ignores the support everyone received.

## How much do individual results vary around the average?

A lot. The prescribing information reports how many people crossed each threshold, and those figures show the spread an average hides.

| SURMOUNT-1, week 72, lowest studied dose | Share of group |
|---|---|
| Average weight change | −15.0% |
| Lost at least 5% | 85.1% |
| Lost at least 10% | 68.5% |
| Lost at least 15% | 48.0% |
| Lost at least 20% | 30.0% |
| Placebo, lost at least 5% | 34.5% |

Source: [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b). Percentages are of everyone randomized, with missing data handled by the trial's statistical methods.

Read it as a worked example. With an average of 15.0%, about 15% of the group did not reach a 5% loss, and about half did not reach 15%. Roughly a third of the placebo group reached 5% on counseling alone. [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b) That is the realistic shape of "average 15%": a wide distribution, not 15% for everyone.

## Why can the same trial show two different numbers?

Because the method of counting changes the result. For STEP 1, the journal abstract reports that 86.4% of the semaglutide group lost at least 5%. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) The prescribing information, which analyzes all randomized patients and handles missing data differently, reports 83.5% for the same threshold, and 47.9% at 15% or more. [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b)

Neither is a mistake. They answer slightly different questions. When you see two figures for one trial, look for the footnote: who is in the denominator, and how were people who left the trial handled?

Both trials' main analyses assessed effects regardless of whether people stayed on treatment. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) [2](https://pubmed.ncbi.nlm.nih.gov/35658024/) In SURMOUNT-1, 14.3% to 16.4% of people in the tirzepatide groups stopped the study drug, compared with 26.4% on placebo. [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b) Weight at week 72 was missing for 9.4% to 10.5% of the tirzepatide groups and 21.6% of the placebo group. [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b)

## Who was enrolled, and are they like you?

Partly. Both trials enrolled adults with obesity, or with overweight and at least one weight-related condition, and excluded people with type 2 diabetes. [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b) [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b)

| | Semaglutide trial (STEP 1) | Tirzepatide trial (SURMOUNT-1) |
|---|---|---|
| People randomized | 1,961 | 2,539 |
| Length | 68 weeks | 72 weeks |
| Mean age | 46 (range 18–86) | 45 (range 18–84) |
| Female | 74% | 68% |
| Mean starting weight / BMI | 105.3 kg / 37.9 | 104.8 kg / 38 |
| Included people with type 2 diabetes | No | No |

Sources: [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b) [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b).

Women in their mid-forties were well represented, which is useful. But the sources cited here do not report how many participants were perimenopausal or postmenopausal, so they cannot tell you whether results differ across menopause. The average starting BMI was also high, near 38. People with type 2 diabetes lost less on average in the same programs: −9.6% versus −3.4% on placebo in the Wegovy diabetes trial, and −12.8% to −14.7% in the Zepbound diabetes trial. [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b) [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b)

## Do trial results apply to compounded products?

No. These trials tested the FDA-approved, branded products. [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b) [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b) A compounded preparation is not FDA-approved, and evidence for a branded medicine does not transfer to it on safety, effectiveness, or quality. Our explainer on [compounded versus brand-name GLP-1 medications](https://joinelovi.com/blog/compounded-vs-brand-name-glp-1) covers the difference. Other trials, including ones of an oral tablet and a higher injectable dose, are described in the same prescribing information, and their numbers should not be read through the figures on this page. [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b)

## A checklist for reading any weight-loss headline

Before you trust a number, check:

| Question | Where to find it | Why it matters |
|---|---|---|
| What was the placebo group's result? | Results table | The gap, not the headline, is the medicine's attributed effect. |
| How long did it run, and when was it measured? | Methods | A week-68 result is not a month-three result. |
| Who was enrolled? | Baseline table | Age, starting BMI, and diabetes status change the average. |
| Was lifestyle support given to everyone? | Methods | Counseling is part of what was tested. |
| Were people who stopped counted? | Footnotes | Counting only completers can look better. |
| Which product, and which dose group? | Title and methods | Dose groups differ; compounded products were not tested. |
| Who paid for the trial? | Funding line | Both trials were funded by the drug makers. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) [2](https://pubmed.ncbi.nlm.nih.gov/35658024/) |

## What can you ask your clinician?

Bring the checklist and ask: Which trial group looks most like me, and where might I differ? What would a reasonable check-in point be for me? What would make you want to reassess my plan? Our guide to [preparing for your follow-up appointment](https://joinelovi.com/blog/preparing-for-your-glp-1-follow-up-appointment) helps you organize notes, and the article on [why weight loss may be slower than expected](https://joinelovi.com/blog/why-weight-loss-on-a-glp-1-may-be-slower-than-expected) covers the timing question in detail.

If you feel unwell on treatment, contact your prescriber. This page cannot tell you whether a symptom is expected for you.

## Common questions

### Does a 15% average mean I will lose 15%?

No. It is the group's average. In the example above, about half the group did not reach a 15% loss. [4](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b)

### Is the placebo result "real" weight loss?

Yes, in the sense that people lost weight. Placebo participants received lifestyle counseling too, which is why the difference between groups is the useful number. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) [3](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b)

### Can I compare the two trials directly to say which drug is better?

Not reliably. They enrolled different people, ran for different lengths, and were not designed to compare the drugs against each other. [1](https://pubmed.ncbi.nlm.nih.gov/33567185/) [2](https://pubmed.ncbi.nlm.nih.gov/35658024/)

## References

1. [Once-Weekly Semaglutide in Adults with Overweight or Obesity \(STEP 1\)](https://pubmed.ncbi.nlm.nih.gov/33567185/) — The New England Journal of Medicine
2. [Tirzepatide Once Weekly for the Treatment of Obesity \(SURMOUNT-1\)](https://pubmed.ncbi.nlm.nih.gov/35658024/) — The New England Journal of Medicine
3. [WEGOVY \(semaglutide\) prescribing information, section 14 Clinical Studies](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b) — DailyMed, National Library of Medicine
4. [ZEPBOUND \(tirzepatide\) prescribing information, section 14 Clinical Studies](https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b) — DailyMed, National Library of Medicine

For general education; this does not replace medical advice or an individual evaluation. Talk to your clinician before starting, changing, or stopping treatment. Compounded medications are not FDA-approved.