How many people stopped retatrutide in the trials?

The single most useful safety number in a trial is how many people stopped the drug because of side effects. For retatrutide it ranged from about one in fifty on the lowest dose in a diabetes trial to nearly one in five on the top dose in TRIUMPH-4. This page explains what that figure measures, why placebo groups have drop-outs too, and why a trial rate is the floor for what to expect outside one.

The numbers

All figures are the share of people in each arm who stopped the study drug because of an adverse event. Arm labels are trial design facts; see TRIUMPH-1 for the trial itself.

What "discontinuation due to adverse events" means

Trials count several kinds of stopping. People leave because they move house, lose interest, get pregnant, or simply cannot face another 80 weeks of visits. Discontinuation due to adverse events is the narrower figure: participants who stopped the study drug because of a medical event, whether or not the drug caused it. It is the best single measure of how many people found treatment intolerable, because it captures every reason in one number, from relentless nausea to a serious illness.

It is not a measure of how many people had side effects. In TRIUMPH-1, 42.4% of people on the top dose had nausea, but only 11.3% stopped for any adverse reason.^[1] Most people who felt sick stayed the course.

Why placebo groups have drop-outs

Around one in twenty people on dummy injections also stopped because of an adverse event. That is normal. Over 80 weeks people get ill for reasons unrelated to the trial, and some of those illnesses lead them to withdraw. A sixth of placebo participants in TRIUMPH-1 reported nausea, because expectation, diet changes and injecting anything at all have effects.^[1] The placebo rate is the background noise. The drug's true contribution is the gap between the two columns.

Why the 4 mg rate can be below placebo

In TRIUMPH-1 and TRIUMPH-2, fewer people stopped on the lowest dose (4.1% and 3.8%) than on placebo (4.9%).^[1][3] That does not mean the low dose caused fewer problems than a saline injection. With a few hundred people per arm, a difference of under one percentage point is chance. It may also reflect something real: people losing a lot of weight and feeling their health improve may be more motivated to put up with side effects and stay in. Either way, "below placebo" means "not measurably different from placebo", nothing more.

What a one-in-nine rate tells you

11.3% on the top dose of TRIUMPH-1 is roughly one person in nine.^[1] Turn that around: eight in nine stayed on the drug for 80 weeks despite the side effects listed on the gut effects and dysesthesia pages. For a drug producing 25% average weight loss, most clinicians would call that acceptable, and analysts did.^[5]

But one in nine is not small. In TRIUMPH-4 it was nearly one in five, and Lilly said some of those people stopped because of "perceived excessive weight loss", mostly among those who started with a lower BMI.^[2] That is a side effect nobody expected to be listing: a drug so effective that some participants felt they were losing too much. In the heavier TRIUMPH-4 sub-group (BMI 35 or more), drop-out rates were noticeably lower.^[2]

Why a trial rate is a floor, not a ceiling

Three things made the trials easier to stay in than unsupervised use could ever be.

1. Supervision. Participants had scheduled visits, blood tests and someone to call about symptoms. Problems were caught early and managed, so fewer became reasons to stop.
2. A known product at a known dose. Every participant received pharmaceutical-grade retatrutide in the amount the protocol specified. Testing of products sold online has found vials at nearly double the labelled strength and others well below it.^[6] Someone injecting from a double-strength vial is not on the 12 mg arm; they are somewhere off the chart.
3. Selection. People with recent heart problems, pancreatitis, significant kidney disease and several other conditions were excluded, as the what we do not know yet page sets out. The people most likely to run into trouble were not counted.

So when someone quotes "only 11% stopped" as reassurance, the honest reading is: 11% stopped under the best possible conditions. Without those conditions, the rate of people who cannot continue, or who continue when they should have stopped, can only be higher. Trial rates describe the trial product. Unlicensed products carry unknown extra risks, and nobody is counting the people who stop because of them.