Definition
A clinical trial is a planned experiment in which a medical intervention (drug, device, procedure) is tested on human subjects to determine its safety and efficacy. The randomised controlled trial (RCT) — random assignment of subjects to treatment and control groups, ideally with blinding — is the gold standard, designed so that probability theory can be applied directly to the resulting data.
Randomisation makes the assignment independent of any unobserved confounders; probability then gives the framework for quantifying whether observed differences are real or chance variation.
Why it matters
How it works
A trial begins with a hypothesis — typically that the new treatment is no better than the current standard (the null hypothesis). Subjects are randomised between treatment and control, given the assigned therapy, and outcomes are measured. Statistical tests then compute the probability of observing the data (or something more extreme) under the null. If this p-value is small enough (conventionally below 0.05), the null is rejected and the treatment is judged effective.
The whole apparatus assumes the probability model is well-specified, which is why trial design, randomisation, blinding, and pre-registration matter so much: they keep the assumptions defensible.