Randomised controlled trial
- Study design
Chance decides who gets the intervention and who gets the control. The groups end up alike, so the final difference is put down to the intervention.
What it is
In a randomised controlled trial (RCT) the investigator assigns the intervention at random. Randomisation spreads known and unknown factors that affect the outcome evenly between groups, so confounding is controlled on average. That makes it the single design with the lowest risk of bias for evaluating treatments. Its quality depends on allocation concealment, blinding, complete follow-up and intention-to-treat analysis.
How it is built
- The population is defined with inclusion and exclusion criteria.
- A random sequence is generated and allocation is concealed until each patient is enrolled.
- One group receives the intervention and the other the control (placebo or usual care).
- Patients, clinicians and assessors are blinded when possible.
- A pre-specified outcome is measured and analysed by intention to treat.
Worked example
- 600 adults with hypertension in 12 health centres are randomised: 300 get text-message reminders and 300 usual care.
- At baseline the groups are alike: mean age 58 and 57 years; 52% and 51% women.
- At 6 months, blood pressure is controlled in 180 of 300 (60%) with messages and 141 of 300 (47%) with usual care.
- RR = 60% / 47% ≈ 1.28; absolute difference = 60% − 47% = 13 percentage points.
- NNT = 1 / 0.13 ≈ 7.7, rounded up to 8.
- Reading: about 8 people need reminders for 6 months for 1 more to reach blood pressure control.
Strengths and weaknesses
| Strengths | Weaknesses |
|---|---|
| Controls known and unknown confounding | Costly and slow |
| Supports causal inference more securely | Not always ethical or feasible |
| Allows participants to be blinded | Highly selected patients: less generalisable |
| Pre-specified outcomes | Of little use for rare or late adverse effects |
Markers
- RCT
- RR
- NNT
- ITT
When to use it
- It is the design regulators require to approve most new medicines.
- A pragmatic RCT compares strategies in real conditions; an explanatory RCT tests efficacy in ideal conditions.
- When the intervention is applied to groups (schools, municipalities), a cluster-randomised trial is used.
Common pitfalls
- Confusing randomisation (how the sequence was generated) with random sampling (how the population was chosen).
- Dropping from the analysis people who did not adhere: it breaks the balance chance created.
- Believing randomisation guarantees identical groups in small trials: chance can leave imbalances.
Test yourself
Think of the answer, then tap the question to see it.
What is the main advantage of randomisation?
It spreads confounders evenly between groups, both the known and the unknown ones.
In an RCT, the control group reaches the outcome in 30% and the intervention group in 40%. What is the NNT?
The difference is 10 points (0.10); NNT = 1 / 0.10 = 10.
Why is there no RCT of whether smoking causes cancer?
Because it is not ethical to assign people a harmful exposure; observational studies are used instead.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Chapter 8: Assessing risk of bias in a randomized trial (Cochrane, 2024); Principles of Epidemiology | Lesson 1 - Section 7 (Centers for Disease Control and Prevention, 2012).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.