What is a randomized controlled trial?
A randomized controlled trial assigns participants at random to two or more groups: one gets the treatment being tested, the other gets a placebo or existing care, and the outcomes are compared. Because of randomization, age, lifestyle and even unknown factors end up evenly spread between groups, so a difference in outcomes can be attributed to the treatment.
Key safeguards
- Allocation concealment: researchers can't know in advance which group the next participant will join.
- Blinding: participants (single-blind) or both participants and assessors (double-blind) don't know the assignment, reducing expectation bias.
- Placebo control: an identical-looking dummy pill filters out the placebo effect.
- Intention-to-treat (ITT): people who stop treatment are still analyzed in their original group, preserving the benefits of randomization.
Common designs
- Parallel: different people in each group—the most common form.
- Crossover: the same people experience both conditions in sequence, so fewer participants are needed.
- Non-inferiority: tests whether a new treatment is not worse than an existing one by more than a preset margin.
- Pragmatic: designed to resemble real-world care so results generalize more easily.
What are the limitations of RCTs?
RCTs aren't perfect. Strict eligibility criteria can make participants unlike real patients, hurting external validity, and some treatments—exercise, meditation, surgery—are hard to blind. Trials stopped early for apparent benefit tend to overestimate effects (Bassler et al., 2010), and rare side effects or long-term outcomes are often better captured by observational studies. That's why meta-analyses that pool trials and long-running cohort studies complement them.
FAQ
- Is an RCT always better than an observational study?
- For testing whether a treatment causes an effect, usually yes, but observational studies are better for rare harms, long-term outcomes and questions where randomizing would be unethical.
Sources reviewed: 2026-10-25
References
- Schulz, K. F., Altman, D. G., & Moher, D. (2010). CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. BMJ, 340, c332.
- Bassler, D., et al. (2010). Stopping randomized trials early for benefit and estimation of treatment effects. JAMA, 303(12), 1180–1187.