Placebo-Adjusted Weight Loss Explained: How to Read GLP-1 Trial Results

You probably landed on this page because you saw two different weight-loss numbers for the same drug and wanted to know which one was real. Wegovy at 14.9% in one place. Wegovy at 12.4% “placebo-adjusted” somewhere else. Same trial. Same patients. Two numbers.

Both are correct. Here’s the bottom line.

Placebo-adjusted weight loss is the difference between the average weight loss in the drug group and the average weight loss in the placebo group of a clinical trial. In the Wegovy STEP 1 trial, the drug arm lost 14.9% of body weight at week 68. The placebo arm lost 2.4%. The placebo-adjusted treatment difference listed on the FDA label is −12.4 percentage points. It is the size of the gap between the two groups, not a separate weight loss anyone actually had.

The placebo group did not “do nothing.” They got the same lifestyle counseling, the same study visits, and the same monthly weigh-ins as the drug group. That is why they lost weight at all — and why the placebo-adjusted number is the one regulators care about, but also why it is not the number you will likely match in real life.

By the end of this guide you will know exactly what the term means, why the same drug has two or three different “correct” numbers in the same trial, what the placebo-adjusted number is for every FDA-approved GLP-1 weight-loss medication available right now, and which number actually matters for your decision.

The formula is simple:

There is no hidden math. No statistical sleight of hand. It is straight subtraction, with one small wrinkle: regulators use a model-adjusted version called a least-squares mean, which can shift the number by 0.1 or 0.2 percentage points.

You will also see this concept written three other ways. In GLP-1 weight-loss trial reporting, these terms usually point to the same between-group idea:

• Placebo-subtracted weight loss
• Placebo-corrected weight loss
• Estimated treatment difference versus placebo

We will use “placebo-adjusted” throughout.

Why does this number exist at all? Because in any modern obesity trial, both arms lose weight. The placebo arm loses weight from the lifestyle program, the trial environment, and the basic act of being watched. Reporting only the drug arm’s number would credit the molecule for everything that happened to those patients — including the parts that had nothing to do with the drug. The placebo-adjusted number isolates the part that is fairly attributable to the medication itself.

“12 percentage points” is a flat subtraction of two percentages: 15 minus 3. “12% more” is a relative comparison — it would mean the drug arm lost 12% more of what the placebo arm lost, which would be 3.36% body weight. Off by a factor of about four.

When you see a press release that says “Drug X produced 17.8% greater weight loss than placebo,” squint. Almost always, what they mean is 17.8 percentage points of difference — the gap between roughly 21% and 3% in SURMOUNT-1. If a writer actually meant “17.8% more than placebo’s 3%,” the result would be 3.5%, which is not a story anyone would publish.

We use percentage points throughout this guide. When you see “pp” after a number, that is shorthand for percentage points.

The math:

The label rounds slightly differently to −12.4 because the FDA reports a least-squares mean — a model-adjusted average that accounts for missing data and trial structure. This is normal. The underlying meaning is the same: the drug arm lost about 12 to 13 percentage points more body weight than the placebo arm over 68 weeks. (Source: Wegovy prescribing information; Wilding et al., NEJM 2021.)

For someone who started the trial at 232 pounds (the trial’s average baseline), 14.9% works out to roughly 35 pounds lost. The placebo arm averaged about 6 pounds lost. The drug-attributable portion is the 29-pound gap between them.

You can read the placebo-adjusted number as “the part of the weight loss that the drug appears to be doing on top of everything else.” The 6 pounds the placebo arm lost are real pounds — they did not magically come back. But they were not pounds the drug caused.

For a participant at the trial’s average baseline weight of about 231 pounds, those percentages translate to roughly 35, 45, and 48 pounds lost on the three doses. The placebo arm averaged about 7 pounds lost.

The dose-response is real. Higher tirzepatide doses produced bigger losses both in absolute terms and relative to placebo — the kind of internal consistency clinicians and regulators look for as a signal that the molecule is doing real work.

This is the section that most coverage of placebo-adjusted weight loss skips. But the formula only makes sense if you understand what the placebo arm actually went through.

The takeaway: when you see a placebo arm lose 2.4% over 68 weeks, you are looking at the integrated output of those six forces. None of them is the drug. All of them are real. Once you understand them, the placebo-adjusted number becomes much easier to interpret correctly.

This trips up almost every news reader. When a headline says “Wegovy was 12.4% placebo-adjusted,” it does not mean “12.4% of the weight loss came from the placebo effect.” It means “the difference between the two arms was 12.4 percentage points.” The placebo arm’s 2.4% loss includes a small psychological placebo effect, but it also includes real lifestyle change, real behavior shifts, and real attention from clinical staff.

The cleaner way to say it: the placebo-adjusted number is the difference between the two randomized groups. Whatever was happening in the placebo group — drug-related or not — gets subtracted out.

Here is what this looks like in practice. SURMOUNT-1 reported two primary numbers for tirzepatide 15 mg at 72 weeks:

• Treatment-regimen estimand: −20.9% body weight loss — includes everyone who was randomized, regardless of whether they stayed on the drug.
• Efficacy estimand: −22.5% body weight loss — includes only people who actually took the drug as prescribed and excludes data points after they stopped.

Same drug. Same trial. Same time point. A 1.6-point gap between two numbers, both of which are correct.

The treatment-regimen estimand answers: What happens on average to people randomized to this drug, across the messy reality of treatment? The efficacy estimand answers: What happens to people who actually take this drug as designed?

The FDA’s E9(R1) guidance, finalized in 2021, gives sponsors a structured framework for defining the treatment effect a trial is designed to estimate — including how intercurrent events like discontinuation are handled — so readers can tell which question the reported number is answering. This is one of the cleanest improvements in obesity-drug reporting of the last decade, and almost no consumer coverage explains it.

For SURMOUNT-5 — the only head-to-head trial of Wegovy vs Zepbound — tirzepatide at maximum tolerated dose produced −20.2% versus semaglutide at −13.7% at 72 weeks. That 6.5 percentage point gap is the most reliable direct comparison available.

The Zepbound SURMOUNT-4 maintenance trial is the cleanest example. Participants who had already lost weight on tirzepatide were randomized to continue or switch to placebo. Those who switched to placebo regained weight rapidly over the next 52 weeks, while those who continued on tirzepatide maintained their loss. The resulting placebo-adjusted gap was dramatically larger than in a standard initiation trial — not because the drug got more effective, but because the placebo arm’s condition changed.

This is why you cannot simply compare placebo-adjusted numbers across different trial types. An initiation trial and a maintenance trial answer completely different questions.

A 2025 JAMA Internal Medicine post-hoc analysis of SURMOUNT-4 found that the cardiometabolic improvements achieved during the active-treatment phase reversed as weight was regained after stopping tirzepatide — reinforcing that the placebo-adjusted gap in a maintenance trial reflects the sustained benefit of continued treatment.

Here is the order to read a trial table:

1. Who was in the trial? Diabetes or no diabetes? Mean baseline BMI? Read the population paragraph before any number.
2. What was the duration? A 12-week result is not comparable to a 72-week result. Long trials often show plateaus; short trials catch the steep part of the loss curve.
3. Find the drug arm’s mean weight change.
4. Find the placebo arm’s mean weight change.
5. Find the estimated treatment difference. If it is not given directly, subtract the two means yourself. If your subtraction doesn’t match exactly, you are looking at LSMean adjustment — normal.
6. Find the responder rates: ≥5%, ≥10%, ≥15%, ≥20%. These tell you how variation distributes around the mean.
7. Find the discontinuation and adverse event rates. Efficacy without tolerability is not useful information.

After that, you can compare. Not before.

If the drug group lost 14.9% and the placebo group lost 2.4%:

To translate to pounds, multiply each percentage by the trial’s baseline weight. For a 232-pound average:

Two caveats:

1. Your subtraction may not match the printed label exactly because regulators report least-squares means. The difference is usually 0.1 to 0.3 percentage points.
2. The trial average is a population average. Your individual result could be substantially higher or lower.