Randomized Controlled Trial (RCT)

A randomized controlled trial is an experimental study design in which participants are randomly assigned to receive either the intervention being tested or a control condition (such as placebo, standard treatment, or no treatment). Random assignment aims to create comparable groups at baseline, allowing differences in outcomes to be attributed to the intervention. RCTs are considered the gold standard for evaluating the efficacy of medical interventions, and they provide key safety data, although rare or long term harms often also need observational evidence.

Among individual study designs, the RCT occupies the highest position in the hierarchy of evidence for questions about treatment effectiveness because it provides the strongest protection against confounding and selection bias. Drug regulatory agencies, including the FDA and EMA, generally expect evidence from adequate and well controlled trials, usually randomized, before approving new treatments; the FDA's standard (21 CFR 314.126) also accepts other controls, such as historical controls, in special circumstances, for example diseases with high and predictable mortality. Clinical practice guidelines rely heavily on RCT evidence. Systematic reviews and meta analyses of RCTs provide the most reliable estimates of treatment effects and inform decisions that affect millions of patients worldwide.

The RCT process begins with formulating the research question, typically structured around PICO (population, intervention, comparator, outcome). The study protocol specifies eligibility criteria, the intervention and comparator, primary and secondary outcomes, sample size (based on power analysis), randomization procedures, blinding strategy, and the statistical analysis plan.

Participants who meet the eligibility criteria and provide informed consent are randomly assigned to the treatment or control group. Randomization can be simple, blocked, stratified, or adaptive (such as minimization). Allocation concealment prevents those enrolling participants from knowing upcoming assignments. Blinding, when feasible, prevents participants, caregivers, outcome assessors, or data analysts from knowing group assignments, reducing performance and detection bias.

During the trial, participants receive the assigned intervention or control, and outcomes are measured at prespecified time points. An intention to treat analysis includes all participants in their assigned groups regardless of adherence, preserving the randomization benefit. Per protocol analysis, which includes only adherent participants, may be reported as a secondary analysis but is susceptible to bias from differential adherence.

Trials of investigational products are conducted under Good Clinical Practice (GCP), set out in the ICH E6 guideline (currently E6(R3), adopted by ICH in January 2025), which specifies requirements for ethics approval, informed consent, data management, adverse event reporting, and monitoring; its principles may also be applied to other interventional trials. In many trials, especially large ones or those in serious conditions, an independent data safety monitoring board (DSMB) reviews interim analyses to identify early evidence of benefit, harm, or futility that may warrant stopping the trial.

RCTs can follow parallel group, crossover, factorial, or cluster randomized designs, depending on the research question and practical constraints. Each design has specific analytical requirements and assumptions. Trial reports should follow the CONSORT 2025 statement, a 30 item checklist with a participant flow diagram that superseded CONSORT 2010, and trial protocols the SPIRIT 2025 statement.

Use an RCT when the primary research question concerns the efficacy or safety of an intervention and when randomization is ethically and practically feasible. RCTs are the design of choice for evaluating new drugs, devices, surgical procedures, behavioral interventions, and health system innovations. They are less suitable for evaluating exposures that cannot be ethically manipulated (such as smoking or pollution), for rare outcomes that require very large samples, or for questions about long term outcomes when follow up is infeasible. In these situations, observational designs with appropriate analytical methods may be more appropriate.

A frequent error is failing to conceal allocation adequately, allowing enrollment staff to predict or influence group assignments. Another mistake is unblinding participants or assessors unnecessarily, introducing performance and detection bias. Researchers sometimes analyze only participants who completed the protocol (per protocol analysis) as the primary analysis, abandoning the intention to treat principle and potentially biasing results. Reporting only statistically significant outcomes while omitting null results for prespecified secondary outcomes is a form of selective reporting that distorts the evidence. Finally, generalizing RCT results to populations that differ substantially from the trial sample requires caution, as the controlled conditions of a trial may not reflect real world practice.

A double blind, placebo controlled RCT enrolls 600 adults with moderate to severe eczema, randomly assigned 1:1 to receive a new biologic drug or placebo by subcutaneous injection every two weeks for 16 weeks. The primary outcome is the proportion achieving at least 75% improvement in the Eczema Area and Severity Index (EASI 75) at week 16. Intention to treat analysis shows EASI 75 is achieved by 38% of the treatment group versus 12% of the placebo group (risk difference = 26 percentage points, 95% CI: 19 to 33, p < 0.001). Adverse events are similar between groups. The trial is registered on ClinicalTrials.gov, and the full protocol and statistical analysis plan are published as supplements.