Key Takeaways
- A phase describes the main development question for a drug or biologic; it is not a quality score, approval status or prediction of personal benefit.
- Phase I emphasises dose, pharmacology and safety; Phase II looks for preliminary activity; Phase III usually confirms benefit–risk against a comparator; Phase IV occurs after marketing approval.
- Phase 0, Phase I/II, Phase II/III, dose-expansion and adaptive designs do not fit a simple staircase but should still state a clear objective for each part.
- Devices, behavioural interventions and many observational studies may correctly show “not applicable” rather than a phase.
- The consent form must explain the experimental elements, assignment, procedures, risks, alternatives, costs and the possibility of no direct benefit.
Content
Trial “phase” is a shorthand for the question a study is designed to answer at a point in medical-product development. It is useful, but incomplete. Two Phase II trials can differ in randomisation, population, comparator, endpoint, maturity and risk. A patient should therefore read the phase together with the protocol title, intervention, eligibility, arms, primary endpoint, recruitment status and earlier data.
ClinicalTrials.gov defines phases for drug and biological-product studies and uses “not applicable” for studies such as device or behavioural trials that do not follow FDA drug-phase definitions [1]. An observational registry also should not be made to look like a treatment trial by assigning it a phase.
The development path before human trials
Discovery and preclinical work examine the product in laboratories and, where relevant, animals. They explore pharmacology, manufacturing and potential toxicity. Human testing begins only after the applicable regulatory and ethics processes permit it; “successful in mice” is not a clinical phase and does not establish safety or effectiveness in people. FDA’s development overview separates discovery, preclinical research, clinical research, regulatory review and post-market safety monitoring [2].
What each phase is mainly trying to learn
Phase · Main question · Typical design features · What a patient should not assume
Early Phase I / Phase 0 · Does the product reach or affect the intended target in humans? · Very limited exposure or microdosing; exploratory sampling · It is intended to treat the disease or provide benefit
Phase I · How should it be given, at what dose, and with what early safety/pharmacology? · Small cohorts, dose escalation, food or interaction studies; oncology often enrols patients rather than healthy volunteers · The “highest dose” is best, or tumour response is the main proof
Phase II · Is there enough preliminary activity and safety to justify further testing? · Single-arm or randomised; disease-specific cohort; response or surrogate endpoints may be used · A promising response rate proves superiority or approval
Phase III · Does benefit–risk compare favourably with standard care or another control in the intended population? · Often randomised and multicentre; predefined primary endpoint and statistical plan · Every participant receives the new treatment, or the new arm is known to be better
Phase IV · What else can be learned after approval? · Post-market safety, effectiveness, optimal use, new populations or required commitments · “Post-market” means no remaining uncertainty or no research procedures
NCI describes the same broad progression for cancer treatment trials: early phases focus on safety, dose and signals of effect; later trials compare with current standard treatments, and Phase IV examines longer-term safety and effectiveness after approval [3]. Participant numbers often grow across phases, but size alone does not define the phase.
Early Phase I or Phase 0: exploration, not treatment
ClinicalTrials.gov describes Early Phase I (formerly Phase 0) as very limited human exposure with no therapeutic or diagnostic goal [1]. A microdose study may show whether a drug reaches a tumour or affects a biomarker without using a dose expected to treat disease.
Ask how much product is given, whether repeat dosing occurs, what biopsies or imaging are required, and whether any direct clinical benefit is expected. The consent form should not use the emotional appeal of access to an experimental product to blur a nontherapeutic purpose.
Phase I: dose and safety come first
Phase I may evaluate route, schedule, metabolism, pharmacokinetics, pharmacodynamics and dose-limiting toxicities. In many non-oncology studies, healthy volunteers are enrolled. Cancer Phase I trials commonly enrol patients because exposing healthy people to cytotoxic or high-risk anticancer products would be inappropriate.
A dose-escalation design may start small cohorts at a lower dose and open higher cohorts after safety review. Participants do not choose the dose simply because a higher one sounds stronger. The recommended dose for later study is not necessarily the maximum tolerated dose; it may balance activity, exposure, chronic toxicity and feasibility.
Expansion cohorts can enrol more people at a selected dose, sometimes in a biomarker or tumour-defined group. Response observations are important but can be unstable in small, selected cohorts without a comparator. Ask which results came from escalation and which from expansion, how many patients had the same disease and biomarker, and how long they were followed.
Phase II: a signal is not confirmation
Phase II asks whether the intervention shows sufficient preliminary effectiveness in a defined disease and continues safety evaluation. It may be single-arm, compare against historical outcomes, or randomise between treatments. The primary endpoint could be response rate, progression-free survival, symptom change, a laboratory measure or another surrogate.
For a single-arm trial, ask what “response” means, who assesses it, whether scans are independently reviewed, how many participants were evaluable, and what happened to those who discontinued early. A high percentage in a small biomarker-selected cohort cannot be compared casually with a different population.
Randomised Phase II studies may be designed to choose which dose or regimen deserves Phase III, not to provide definitive proof. Read the statistical objective rather than assuming randomisation always means a pivotal trial.
Phase III: comparison depends on the exact question
Phase III commonly compares an investigational strategy with the accepted standard, placebo plus standard care, or another control. Random assignment helps balance known and unknown patient characteristics. Blinding may reduce biased outcome assessment, but not every intervention can be blinded.
Ask whether the trial tests superiority, non-inferiority or equivalence; each asks a different question. Understand the primary endpoint, follow-up duration, crossover rules, stratification factors, interim analyses and whether the endpoint measures survival, function, symptoms or a surrogate. A “positive” trial means the prespecified analysis met its criteria; it does not mean every patient benefited or that harms were unimportant.
A placebo is not automatically “no care.” In serious diseases it may be added to standard therapy, used when no proven treatment exists, or used for a limited component. The consent form should state exactly what each arm receives and the probability of assignment.
Phase IV: approval is a milestone, not the end of evidence
Phase IV studies occur after a product is approved for marketing and may investigate rarer or longer-term harms, real-world effectiveness, optimal use, interactions, new populations or regulatory commitments [1]. Verify whether the product is approved for the same disease, biomarker, dose and combination being studied; an approved drug can still be experimental in a new use.
Routine care databases and pharmacovigilance studies may contribute post-market evidence without being interventional Phase IV trials. Conversely, a Phase IV label does not remove study-specific procedures, randomisation or consent.
Why phases are sometimes combined
Phase I/II and Phase II/III trials use one protocol to move between objectives when prespecified criteria are met. Adaptive designs may modify allocation, sample size, doses or cohorts according to planned rules. “Seamless” does not mean oversight is skipped. Ask:
- where the transition occurs and who reviews it;
- whether the dose or population changes;
- which data belong to each objective;
- whether participants are re-consented when important new information appears;
- whether an independent data-monitoring committee is involved.
The phase concept describes development and is not a requirement that every programme follow identical steps. Regulatory approval may sometimes rely on evidence generated in designs that do not resemble a classic large Phase III sequence; approval status must be verified separately from phase.
Phase does not answer the personal decision
For the actual study, determine:
- Purpose: What is the primary scientific question?
- Prior evidence: What human data exist for this exact product, disease, dose and combination?
- Assignment: How many arms are there, and what is the chance of each?
- Standard care: What would the patient receive outside the study, and is any usual option being delayed or excluded?
- Procedures: Which scans, biopsies, admissions, washouts and contraceptive requirements are research-only?
- Endpoint: What counts as success, who measures it and when?
- Risk: Which toxicities are known, suspected and still unknown?
- Practical burden: Who pays for the investigational product, standard treatment, tests, travel, accommodation and care for complications?
- After the trial: Is continued access possible, and who provides follow-up if the patient returns home?
Phase is only one line in this assessment.
Consent and participant protection in China
China’s 2020 Good Clinical Practice rules place participant rights and safety before scientific and social benefit, require ethics review and written informed consent, and require information on purpose, randomisation, procedures, experimental aspects, risks, expected benefit and the possibility of no benefit, alternatives, compensation, costs and the right to withdraw [4]. A signed copy should be provided to the participant.
China published a revised drug GCP in June 2026 that is scheduled to take effect on 1 September 2026, replacing the 2020 version [5]. For a study spanning that date, ask the site which approved documents and transition arrangements apply. Regardless of the version, consent is a continuing conversation: material new information should be communicated, and participation remains voluntary.
An interpreter must translate the clinical meaning, not merely help with registration. Verify that the ethics-approved consent language is understandable, that questions can be answered by the investigator, and that emergency contacts work outside office hours.
Expanded access is not another trial phase
Expanded or compassionate access is a separate pathway for possible treatment use outside a clinical trial in some serious or life-threatening conditions. It is not Phase V and does not guarantee supply or benefit. FDA explains that the treating physician, company, ethics board and regulator may all need to agree, and the investigational product may be ineffective or cause unexpected serious harm [6]. China has its own laws and institutional processes; do not assume a foreign expanded-access pathway applies there.
Medical disclaimer: This guide is general educational information, not trial eligibility or treatment advice. Discuss the protocol and alternatives with the investigator and an independent clinician who knows the complete history. Urgent symptoms require local medical care.
FAQ
Does a higher trial phase mean the treatment is better?
No. A later phase usually means a later development question and more accumulated information, not that the intervention is personally better or safer. Compare the exact evidence, comparator, endpoint and patient population.
Can Phase I participants receive a dose too low to help?
Yes. Dose-escalation studies may assign lower doses to learn about safety and exposure, and benefit may be uncertain at every dose. The consent discussion should explain the assigned cohort and therapeutic uncertainty.
Does Phase III guarantee that I will receive the new treatment?
No. Many Phase III trials randomise participants between an investigational and control arm. The number of arms and probability of assignment should be stated clearly before consent.
Why does a device or behavioural study say “phase not applicable”?
The I–IV phase system was developed for drugs and biological products. Device, behavioural and some other studies use different development and evaluation frameworks, so “not applicable” can be correct.
Is a Phase IV trial ordinary treatment?
Not necessarily. The product is marketed, but the study may test a new population, use or strategy and may include randomisation, extra procedures and research data collection. Read the protocol and consent form.
Sources
- ClinicalTrials.gov — Glossary: Clinical Trial Phases
- US Food and Drug Administration — The Drug Development Process
- US National Cancer Institute — How Clinical Trials Work
- National Medical Products Administration and National Health Commission of China — Good Clinical Practice for Drug Trials, 2020
- National Medical Products Administration, National Health Commission, National Administration of Traditional Chinese Medicine and National Disease Control and Prevention Administration — Revised Drug GCP, Effective 1 September 2026
- US Food and Drug Administration — Expanded Access Information for Patients
- US Food and Drug Administration — Drug Development and Review Definitions
- US National Cancer Institute — Taking Part in Cancer Treatment Research Studies
Image Review
- Decision: Replaced with a topic-specific ImageGen hero and visually reviewed for medical relevance, obvious generation artifacts and bilingual reuse.
- Editorial note: The image shows a generic conversation in a reception lounge. It contains no phase progression, dose cohort, randomisation, comparison arm, consent document or trial milestone, so it cannot communicate the article’s subject.