When a page on this site says a compound is "mostly preclinical" or "in phase 3 trials", it is describing where the compound sits in a process that exists specifically to find out whether something is safe. Knowing the stages makes that shorthand readable.
The phases
| Phase | Who | Size | What it answers |
|---|---|---|---|
| Preclinical | Cells, then animals | — | Is there a plausible mechanism, and does it kill animals? |
| Phase 1 | Usually healthy volunteers | 20–100 | What does the body do with it? What dose is tolerable? |
| Phase 2 | People with the condition | 100–500 | Does it appear to work? What dose should phase 3 use? |
| Phase 3 | People with the condition | Hundreds to thousands | Does it work against placebo, and what are the side effects? |
| Phase 4 | Everyone prescribed it | Ongoing | What shows up once millions of people use it? |
Most compounds die between phases. That is the system working: the failures are the point, not a flaw.
Why the placebo group matters so much
People improve for reasons unrelated to treatment. Conditions fluctuate, people change their behaviour when enrolled in a study, and reporting changes when someone believes they are being treated.
In retatrutide’s TRIUMPH-1 trial, the placebo group lost 2.2% of body weight over 80 weeks. Without that comparison, the drug groups’ results would be impossible to interpret.
How side effects get counted
Published trials report adverse events as rates in each group, including placebo. That comparison is the entire value of the number. Nausea in 42% of people on a drug sounds alarming until you see 14.8% on placebo, and then it tells you something specific.
- Discontinuation rate is often the most honest single figure: how many people stopped because of side effects.
- Serious adverse events are tracked separately from common ones.
- Boxed warnings come from findings serious enough to appear at the very top of the label.
What accelerated approval means
Some drugs are approved on an intermediate measure rather than a final outcome, with confirmatory evidence required afterwards. Elamipretide (SS-31) was granted accelerated approval for Barth syndrome in September 2025 on the basis of knee extensor strength, after it did not beat placebo on its primary endpoints in the controlled portion of the study.
Accelerated approval is a real approval, but it signals that the evidence is thinner than a standard approval. The distinction is worth noticing on any page that mentions it.
What none of this covers
A compound that has never entered this process has no safety profile. Not a clean one, not a worrying one: none. That is a different situation from a drug with known side effects, and it is the situation for most compounds sold as research peptides.