Randomised Controlled Trials

Randomised controlled trials (RCTs) are the gold standard for evaluating the effects of interventions in medicine. Their defining feature is random allocation of participants to intervention or control groups, which distributes both known and unknown confounders evenly between arms, enabling causal inference.

A clinical trial is an experiment on patients designed to evaluate the benefits and risks of one or more treatments. Trials progress through phases: Phase I (safety, dose-finding in a small healthy sample), Phase II (preliminary efficacy and safety), Phase III (definitive efficacy and safety in a large patient population), and Phase IV (post-marketing surveillance).

Primary outcomes are the principal measures of treatment effect. These may be clinical outcomes (mortality, hospitalisation) or surrogate variables — biological measures that capture the effect of treatment on the disease process and that must fully capture the effect on the clinical outcome. Biomarkers (CD4 count in HIV, LDL-cholesterol) may be useful primary endpoints at Phase II but require validation as surrogates before use at Phase III.

Research questions are framed as superiority (does the new treatment outperform the control?) or non-inferiority (is the new treatment no worse than the existing standard?).

Randomisation aims to create groups with the same distribution of baseline characteristics. Concealment of allocation prevents the recruiter from predicting which group the next participant will enter — this is separate from blinding. Blinding prevents participants, clinicians, and outcome assessors from knowing group allocation, minimising performance bias, detection bias, and placebo effects.

Intention-to-treat (ITT) analysis analyses participants as randomised, regardless of what they actually did during the trial. This preserves the benefits of randomisation, controls confounding, and reflects real-world effectiveness. Per-protocol analysis evaluates only participants who fully adhered — it is closer to an estimate of efficacy but reintroduces confounding.

Efficacy describes whether a treatment works under ideal, controlled conditions. Effectiveness describes how well a treatment performs in real-world practice where non-adherence and other factors operate.

Internal validity in an RCT is threatened by inadequate allocation concealment, lack of blinding, loss to follow-up, and non-adherence. Evidence-based practice (EBP) integrates the best available evidence, clinical expertise, and patient values and preferences to determine optimal treatment, recognising that good evidence prevents adoption of ineffective or unnecessarily expensive treatments.