Placebo-Controlled GLP-1 Trials: What They Tell Us (and Don't)

A placebo-controlled GLP-1 trial randomly gives some people the real medicine and others an identical-looking dummy, then compares what happens. It tells you how much of the result came from the medicine itself rather than from expectations, extra support, or time passing. What it can't tell you is how the drug compares with other drugs, how it works in everyday life, or what happens far beyond the trial's end date.

Here's how to read one, with a lot of respect for the unsung heroes of every trial: the placebo group.

What does "placebo-controlled" actually mean?

Most big GLP-1 trials follow the same recipe:

• Randomized: a computer decides who gets the drug and who gets placebo, so the groups start out alike on average.
• Double-blind: neither participants nor the study staff assessing them know who got what.
• Placebo: an inactive injection or pill that looks the same as the real thing.
• Same extras for everyone: both groups usually get the same diet and activity counseling and visit schedule.

So the only planned difference between the groups is the medicine. Whatever gap opens up between them is the best estimate of what the drug added.

Why do placebo groups lose weight?

This surprises people. The placebo group is not sitting still:

Trial	Placebo group's average weight change
STEP 1 (semaglutide, 68 weeks)	−2.4%
SURMOUNT-1 (tirzepatide, 72 weeks)	−3.1%
ATTAIN-1 (orforglipron, 72 weeks)	−2.1%

And some placebo participants did very well. In STEP 1, 31.5% of the placebo group lost at least 5% of their weight. In SURMOUNT-1, 35% did.

Why? They got lifestyle counseling, regular weigh-ins and the general "someone is watching" effect of being in a study. That's precisely why the placebo group exists. Without it, you might credit the drug for weight loss that the counseling and attention produced.

The headline lesson: the drug's real effect is the difference. In STEP 1 that was −14.9% minus −2.4%, a treatment difference of about 12.4 percentage points.

Why do placebo groups report side effects?

Here is a fun one. In the STEP 4 trial, after the randomized phase began, 26.1% of people on placebo reported gastrointestinal events, compared with 49.1% of those who continued semaglutide.

Some of that is ordinary life: people get stomach bugs, eat a bad burrito, have stressful weeks. Some of it is the nocebo effect, where expecting side effects makes you notice or feel them. Placebo groups also stop treatment because of adverse events: 2.6% in SURMOUNT-1, 2.7% in ATTAIN-1, and 8.2% over roughly three years in SELECT.

This is why a side effect's rate in the drug group means little alone. What matters is how much higher it is than in the placebo group.

Spot the trick

1. "Half of trial participants had nausea, so the drug causes nausea in half of people."

Show answer

Check the placebo row. If a chunk of the placebo group also reported nausea, the drug's added effect is the difference between the groups, not the total.

2. "Even the placebo group lost weight, so the drug barely matters."

Show answer

The placebo group got counseling and support, which helps some people. The question is how much more the drug group lost. In SURMOUNT-1 that gap was large: −20.9% on the 15 mg dose vs −3.1% on placebo.

3. "This drug beat placebo, so it beats the other GLP-1 too."

Show answer

Beating placebo says nothing about beating another drug. That takes a head-to-head trial. SURMOUNT-5 did exactly that, comparing tirzepatide with semaglutide (−20.2% vs −13.7% at 72 weeks), and it was open-label, meaning people knew which drug they got.

Scoring: 3 out of 3 means you'd make a great trial monitor. Fewer means you just learned something, which is the whole point.

Is placebo always given from the start?

Not always. Some trials use a clever twist called a randomized withdrawal design. In STEP 4, all 902 participants took semaglutide for 20 weeks first. The 803 who reached the full 2.4 mg dose were then randomly assigned to keep going or switch to placebo for 48 more weeks.

Those who continued semaglutide lost another 7.9% on average. Those switched to placebo regained 6.9%. That design answers a different question ("what happens if you stop?") than STEP 1 ("does it work compared with nothing?"). For the long version, see what happens when you stop semaglutide.

What can't a placebo-controlled trial tell you?

A strong placebo trial answers its own question very well. It leaves other questions open:

• How does it compare with other medicines? Needs a head-to-head trial.
• How does it work in everyday care? Trial participants get frequent visits, free medicine and close follow-up. Real life has copays, shortages and busy schedules. That's where real-world data comes in.
• Who wasn't included? Exclusion criteria shape who the results apply to. Our post on who gets left out of GLP-1 trials digs in.
• What about rare or long-term effects? A 68-week trial of 2,000 people may not show something that happens to 1 in 10,000, or after five years.
• Is it perfectly blind? Noticeable side effects like nausea could hint to some people which group they're in. Researchers can't fully rule this out, which is one reason objective outcomes (measured weight, heart attacks) carry more weight than self-reported ones.

Why do placebo trials matter so much for safety?

Because they let regulators separate "happens on the drug" from "happens anyway." In January 2026, when FDA asked for the suicidal-ideation warning to be removed from Wegovy, Zepbound and Saxenda labels, one key piece of evidence was a meta-analysis of 91 placebo-controlled GLP-1 trials with 107,910 patients. It did not show an increased risk compared with placebo. Individual reports had raised the question; placebo comparisons helped answer it.

Placebo-controlled trials are also a big part of how medicines move through the trial phases toward approval. When you read one, the most useful habit is simple: find the placebo column first, then look at the drug.

Common questions

Is it ethical to give placebo in GLP-1 trials?

Placebo groups in obesity trials usually still receive lifestyle counseling and medical monitoring, and participants agree to the chance of getting placebo through informed consent. Ethics committees review each trial's design before it starts.

How much weight do people lose on placebo in GLP-1 trials?

In STEP 1, SURMOUNT-1 and ATTAIN-1, placebo groups lost about 2% to 3% on average over 68 to 72 weeks, and roughly a third lost at least 5%.

Does double-blind mean nobody can guess their group?

It means the treatments look identical and nobody is told. Some people may still guess based on side effects or how much weight they lose, which is a known limitation.

Does a placebo-controlled trial prove a drug is safe?

It shows which side effects happen more often than with placebo over the trial period. Rare or long-term effects may need larger, longer studies and ongoing safety monitoring after approval.