From Phase 2 to Phase 3: The Go or No-Go That Decides a Drug Program

The move from phase 2 to phase 3 is a commitment decision, and it turns on three questions asked together: did phase 2 find a dose worth carrying forward, is the benefit large enough and steady enough to be real, and does the safety picture leave room for that benefit to matter. A confirmatory phase 3 trial is the largest and most expensive test in the program, so the transition is less about proving success than about deciding whether the case can carry a definitive study. The honest version of this decision is willing to say no. This article is general education about how drug programs are run, not medical advice, and any question about a specific treatment belongs with a clinician who knows your history.

Phase 2 is where a program learns: it searches for the right dose and a first credible signal that the drug does what it should, at a size worth having. Phase 3 is where a program confirms, testing the chosen dose and endpoint at scale, in a population close to real use, against a fair comparison, long enough to hold up to regulatory scrutiny. The transition is where learning must convert into conviction.

What phase 2 is actually trying to deliver

Phase 2 has a narrow job that is easy to blur. It is not there to prove the drug works in the way an approval requires; its task is to reduce the biggest uncertainties cheaply, so the expensive study is aimed correctly.

The first deliverable is a dose. By testing several doses against control, a well-built phase 2 traces the shape of the response and finds where benefit is high while side effects stay acceptable. Choosing too few doses, or spacing them badly, is one quiet way a program reaches phase 3 without knowing which amount to carry. Next comes an effect large enough and consistent enough to be believed, on an endpoint still defensible at full scale. Third is a safety readout clean enough to justify exposing far more people for far longer.

I saw this discipline up close during my years in global drug development at Novo Nordisk, where I worked as an international medical manager on GLP-1, insulin, and combination therapies. The recurring lesson was that phase 2 exists to buy information, and a program that treats it as a dress rehearsal for success rather than a genuine test of assumptions tends to pay for it later.

The three things the transition decision weighs

When a team sits down to decide go or no-go, the evidence sorts into three buckets, and all three must hold at once.

Dose comes first. The transition needs a defensible answer to a plain question: which dose, and why this one rather than a neighbor. If phase 2 leaves the dose ambiguous, phase 3 inherits the ambiguity, and a confirmatory trial is a costly place to still be guessing.

Effect size comes second. A result that is statistically significant in a small study can still be too small to matter clinically, or too fragile to survive a larger, more varied population. What matters is whether the drug beat control by a margin wide enough to stay meaningful as the sample grows and the setting gets messier.

Safety comes third, and it is often decisive. Phase 2 is usually too small to rule out uncommon harms, so the transition reads the safety signal for direction and plausibility rather than final proof. A benefit that looks real can still be outweighed by a risk that could grow with exposure. Benefit and risk are judged together, never on separate ledgers.

Why so many programs fail at exactly this step

Phase 3 is where a striking share of promising programs fall down, and the reasons cluster.

The most common is that the phase 2 signal was real but smaller than it appeared. Fewer participants, shorter follow-up, and selected populations can make an effect look larger and cleaner than it will be in the wider world. When phase 3 recruits a broader group and follows them longer, the effect can regress toward something more modest, and modest sometimes falls short of the bar.

A dose carried forward on thin evidence is another, so the confirmatory trial tests an amount that was never quite right. Then there is an endpoint that behaved well in a small setting but does not translate, especially when a short-term marker stood in for the outcome that matters to patients. Add a safety issue that only surfaces once thousands of people are exposed for months. None of these are failures of effort; they are the reason confirmation is required at all, since one learning study is not enough to be sure.

The discipline of an honest go or no-go

The hard part of this decision is not analytical, it is temperament. By the time a team reaches the end of phase 2, a great deal of work, money, and hope has accumulated, and all of it pulls toward go. An honest process builds in resistance to that pull.

Good practice sets the decision criteria in advance, before the data arrive, so the threshold is not quietly lowered to clear whatever the study produced. It reads the full dataset rather than the most encouraging slice, and treats a subgroup that looks good after the fact as a hypothesis to test, not a result to bank. My FDA clinical investigator training reinforced how much of this structure exists to protect the people who enroll: committing thousands of participants to a trial you already doubt is not a neutral act.

A no-go is not a wasted program. Stopping early frees resources for candidates more likely to help patients, and spares participants a trial built on a weak premise. The willingness to walk away is what makes a yes mean something.

What a careful reader can take from this

The phase 2 to phase 3 transition is a filter, and a demanding one by design. A program crosses it when it can name its dose, show an effect both large enough and durable enough to trust, and present a safety picture that leaves that benefit intact, all judged against criteria set in advance.

Read this way, the high failure rate in phase 3 carries some good news. It is partly the system doing its job, catching effects that shrank, doses that were off, and risks that grew, before a drug reaches wide use. The programs that clear the bar honestly are the ones worth the study that follows.