New treatments move through numbered phases before approval, and the numbering describes different questions rather than increasing degrees of confidence in the same one.
Early work happens before humans
Laboratory and animal studies establish basic biological activity and identify obvious toxicity, generating the data required to request permission to begin human testing.
Regulators review that package before any dose is given to a person, and the review covers manufacturing quality alongside safety findings.
Results at this stage translate imperfectly to people, which is a principal reason candidates that looked promising fail once human testing begins.
Animal models reproduce some features of a human disease and not others, so a compound can correct the modeled version of a condition without affecting the illness itself.
The first phase asks about safety and dose
Initial human studies enroll a small number of participants, often healthy volunteers, and escalate doses cautiously while monitoring for adverse effects.
The objective is characterizing tolerability and how the body absorbs, distributes and eliminates the compound, not determining whether it works.
In serious illnesses where healthy volunteers would be inappropriate, these studies enroll patients instead, and the design changes accordingly.
The middle phase looks for a signal
Second-phase studies enroll larger groups with the condition being targeted and look for evidence of activity while continuing to gather safety data.
These trials are often too small to establish effectiveness definitively, and they frequently use surrogate measures rather than the outcomes that matter to patients.
A promising result here is a reason to proceed, not a conclusion, which is why apparent breakthroughs at this stage so often fail to replicate.
The third phase is the decisive test
Large randomized trials compare the treatment against placebo or standard care, with enough participants to detect a difference reliably.
Randomization and blinding exist to prevent expectation and selection from creating an apparent effect, which they demonstrably do in unblinded studies.
Failure at this stage is common, and it usually means the earlier signal reflected small numbers or a surrogate measure rather than genuine benefit.
Monitoring continues after approval
Post-approval surveillance detects rare adverse effects that trials enrolling thousands could not have observed, through reporting systems and required studies.
Labels are revised, warnings added and products withdrawn on the basis of this information, sometimes years after approval.
Anyone considering participation in a trial should discuss it with their own physician, since eligibility, risks and alternatives are specific to the individual.