What Is a Randomized Controlled Trial (RCT)? Definition, Design and Examples

Where you learn this: trial design, randomization and unblinding procedures are covered in the VIARES Clinical Research Associate course.

What is a randomized controlled trial (RCT)?

A randomized controlled trial tests whether an intervention causes a change in an outcome. Participants who meet the inclusion and exclusion criteria give informed consent and are then randomly assigned to one of two or more groups.

The experimental group receives the experimental treatment. The control group receives a placebo, standard of care, another dose or no treatment. Both groups are followed in the same way, and their outcomes are compared at the end of the trial.

ICH E10 explains why the control matters: it shows what would have happened to patients without the test treatment. Without it, a trial cannot separate the drug’s effect from the natural course of the disease, the placebo effect or other care.

The UK spelling is “randomised controlled trial”. In medicine, RCT almost always stands for this design, which many textbooks call the gold standard for testing efficacy.

How do randomized controlled trials reduce bias?

ICH E10 defines bias as the systematic tendency of a trial’s design, conduct, analysis or interpretation to push the estimated treatment effect away from its true value. An RCT uses several layers of protection against it.

• ✓ Randomization: a deliberate element of chance in treatment assignment. According to ICH E9, it tends to balance prognostic factors, known and unknown, between groups, so a confounding variable is unlikely to differ systematically. It also gives a sound basis for statistical inference.
• ✓ Allocation concealment: nobody enrolling a participant can know or predict the next assignment. ICH E9 says the next treatment number is allocated only after entry is confirmed, and details that make assignments predictable, such as block length, stay out of the protocol.
• ✓ Blinding: participants, investigators and assessors do not know who receives what. In a double-blind study the treatments must be indistinguishable from the active treatment in look and taste, which ICH E9 calls the optimal approach.
• ✓ Intention-to-treat: participants are analyzed in the group they were randomized to, whatever their compliance. ICH E9 says preserving the initial randomization in analysis helps prevent bias. See intention-to-treat analysis.

Concealment and blinding are not the same. Concealment protects the assignment before a participant enters the trial; blinding protects it afterwards, during treatment and outcome assessment.

Stratified randomization adds balance for key baseline factors such as disease severity or site. ICH E9 notes that more than two or three stratification factors is rarely necessary.

What are the different types of RCT design?

The trial design is fixed in the protocol before the first participant is randomized. The main configurations are:

• ✓ Parallel-group: each participant is randomized to one arm and stays there. ICH E9 calls it the most common design for confirmatory trials. Each arm is a separate treatment arm.
• ✓ Crossover: each participant is randomized to a sequence of treatments and acts as their own control, often with a washout period between periods. It needs fewer participants but only suits chronic, stable conditions, because carryover can bias the comparison.
• ✓ Factorial: two or more treatments are tested at once. In a 2×2 factorial design, participants get A alone, B alone, both, or neither.
• ✓ Cluster: whole groups, such as clinics or communities, are randomized instead of individuals. Cluster trials have their own CONSORT reporting extension.

RCTs also differ by the question they answer. A superiority trial shows the new treatment is superior to placebo or an active control. A non-inferiority trial shows it is not worse than an active comparator by more than a pre-set margin. ICH E9 notes that a placebo-controlled trial may be unethical in serious illness when a proven treatment exists.

Design choices drive the sample size calculation: the trial must be large enough to detect a clinically meaningful effect size on the primary endpoint and reject the null hypothesis if the treatment works.

RCT vs observational study

An RCT is an experimental (interventional) study: the investigator assigns the intervention. In an observational study, such as a cohort or case-control study, researchers only record what happens to people who chose or received a treatment in routine care.

That difference is why RCTs usually rank near the top of the hierarchy of evidence, below systematic reviews of RCTs. ICH E10 calls the inability to eliminate systematic differences between groups a major problem of studies without a concurrent randomized control.

Observational studies still matter. They can follow far more patients for longer, in real-world conditions, and can answer questions where randomizing would be unethical, such as the harms of smoking.

Limitations and disadvantages of RCTs

• ✓ Generalizability: strict criteria may exclude sicker or older patients. ICH E10 notes that results from such populations can raise questions about real-world effectiveness.
• ✓ Blinding is not always possible: surgery cannot easily be compared with a drug in a blinded way. An open-label study then needs objective endpoints and blinded assessors.
• ✓ Missing data: participants lost to follow-up weaken the intention-to-treat analysis and can affect the results.
• ✓ Size and duration: trials are often too small or too short to detect rare or late adverse effects, and they are expensive and slow to run.
• ✓ Ethics: some questions cannot be randomized, and a placebo group may be inappropriate when effective treatment exists.

Subgroup analyses add another risk: with many comparisons, some will look significant by chance, so planned analyses belong in the statistical analysis plan.

Reporting RCTs: CONSORT and trial registration

CONSORT (Consolidated Standards of Reporting Trials) is the reporting standard for randomized trials. CONSORT 2025 consists of a 30-item checklist and a flow diagram that shows how many participants were enrolled, allocated, followed up and analyzed in each group.

Registration in a trial registry is the second safeguard. The main rules:

• ✓ US law: under FDAAA 801 and 42 CFR 11.24, applicable clinical trials must be registered on ClinicalTrials.gov no later than 21 calendar days after the first participant is enrolled.
• ✓ Journals: ICMJE member journals require registration in a public trial registry at or before first patient enrollment as a condition for publication.
• ✓ WHO: calls registration of all interventional trials a scientific, ethical and moral responsibility. Its ICTRP search portal pools records from primary registries.

Registration lets readers check published results against the original plan. For the Food and Drug Administration, 21 CFR 314.126 lists random assignment, with or without stratification, as the ordinary way to assign patients in an adequate and well-controlled study.

The first randomized controlled trial (1948)

The Medical Research Council trial of streptomycin for pulmonary tuberculosis, published in the BMJ in 1948, is often cited as the first randomized controlled trial in medicine.

The James Lind Library calls its report a methodological landmark. It described concealed random allocation in detail: the allocation schedule was hidden from the clinicians entering patients, so nobody could choose who received streptomycin.

Example from trial practice

Hypothetical example. A double-blind phase 3 trial compares a new drug with placebo at 40 sites. Randomization is stratified by disease severity and run centrally through an interactive response technology (IRT) system.

• ✓ Randomize: once a participant is confirmed eligible, the coordinator enters the severity stratum in IRT. The system returns a kit number, never the treatment name.
• ✓ Monitor: at the next visit the CRA checks that eligibility was confirmed before randomization, that the stratum matches source documents, and that the dispensed kit matches the IRT assignment. A wrong stratum is a protocol deviation, even if the kit was correct.
• ✓ Unblind: if a participant has a serious adverse event and the treating physician needs to know the treatment, the investigator unblinds that one participant in IRT. ICH E6(R3) asks the investigator to document and explain any premature unblinding to the sponsor promptly.

The CRA files the unblinding record and confirms that the rest of the site team, and the other participants’ assignments, stayed blinded.

Where you learn this at VIARES

Section 6 of the VIARES Clinical Research Associate course, “Clinical Trial Design & Clinical Research Roles and Responsibilities”, covers the elements of a study protocol according to ICH GCP, trial design and methodologies. In the assignment, learners explain stratified block randomization, the double-dummy design, the cross-over trial and the factorial design, including when each is used, its restrictions and its bias-reducing effect.

Section 10, “Investigational Medicinal Product Management”, explains the randomization process and its documentation requirements, including un-blinding procedures. It also shows how to monitor randomization of patients and other IP-related aspects during monitoring visits.

In the Clinical Study Coordinator I course, Section 6 “Clinical Research Process and Study Design” covers study designs, randomization and non-randomization, and the phases of clinical research.

FAQ

What does RCT stand for in medicine?

RCT stands for randomized controlled trial, sometimes written as randomized clinical trial. It is a study in which participants are randomly assigned to receive the intervention or a control, so the groups differ only by chance and by the treatment received.

Is a randomized controlled trial experimental or observational?

Experimental. The investigator assigns the intervention by randomization. In an observational study, researchers only observe treatments that people receive in routine care, which leaves more room for confounding.

What was the first randomized controlled trial?

The 1948 Medical Research Council trial of streptomycin for pulmonary tuberculosis is often cited as the first. Its report described concealed random allocation, so clinicians could not know which treatment the next patient would get.

How do randomized controlled trials reduce bias?

Randomization balances known and unknown prognostic factors between groups, allocation concealment stops anyone from steering patients into a group, and blinding keeps knowledge of treatment from influencing care and outcome assessment. Intention-to-treat analysis preserves the randomized groups.

Can an RCT be done without a placebo group?

Yes. The control group can receive an active treatment, standard of care, a different dose or no treatment. An active control is often required when withholding proven treatment would be unethical.

Sources

1. ICH E9: Statistical Principles for Clinical Trials, sections 2.3.1, 2.3.2, 3.1, 3.3 and 5.2 (ICH guideline)
2. ICH E10: Choice of Control Group and Related Issues in Clinical Trials, sections 1.2 and 2.1.7.3 (ICH guideline)
3. ICH E6(R3): Guideline for Good Clinical Practice (2025), section 2.11 Randomisation Procedures and Unblinding
4. 21 CFR 314.126: Adequate and well-controlled studies (eCFR)
5. SPIRIT-CONSORT Group: CONSORT 2025 Statement, 30-item checklist and flow diagram
6. SPIRIT-CONSORT Group: Extensions (cluster, crossover and factorial trials)
7. 42 CFR 11.24: When must clinical trial registration information be submitted? (eCFR)
8. ClinicalTrials.gov: FDAAA 801 and the Final Rule
9. ICMJE: Clinical Trial Registration
10. WHO: International Clinical Trials Registry Platform (ICTRP)
11. James Lind Library: Medical Research Council (1948). Streptomycin treatment of pulmonary tuberculosis. BMJ 1948;2:769–782