Community medicine chapters
Randomised controlled trials and quasi-experiments
How a randomised controlled trial is designed and why randomisation, allocation concealment and blinding matter, intention-to-treat analysis, the phases of drug trials, special trial designs, and quasi-experimental studies when randomisation is not possible.
- In a randomised controlled trial, participants are randomly allocated to intervention or control, which balances known and unknown confounders; it is the gold standard for testing treatments.
- Allocation concealment prevents selection bias at enrolment; blinding (single, double, triple) prevents bias in treatment and assessment; analysis should be by intention to treat.
- Quasi-experiments test interventions without randomisation (for example before-and-after or non-randomised control groups) and are more open to confounding.
On this page
Overview
Experimental (interventional) studies are those in which the investigator assigns the intervention. The best of these is the randomised controlled trial (RCT), the gold standard for judging whether a treatment or preventive measure works.
Steps in an RCT
- Write a protocol: aims, eligibility criteria, intervention, outcomes, sample size, analysis plan. Obtain ethics approval and register the trial.
- Select the study population using inclusion and exclusion criteria, and obtain informed consent.
- Randomise participants to intervention or control.
- Give the intervention (control may be placebo or standard treatment).
- Follow up both groups in the same way.
- Assess outcomes, ideally blinded.
- Analyse (intention to treat) and report (CONSORT).
Randomisation
Randomisation means every participant has a known, equal chance of being in each group, decided by chance (random number tables or computer), not by the investigator or patient.
Why it matters: it makes the groups comparable for both known and unknown factors (confounders), so differences in outcome can be attributed to the intervention. It also removes selection bias.
Types: simple, block (keeps group sizes equal), stratified (balances key factors such as age), and cluster randomisation (groups such as villages or schools).
Randomisation is the only method that controls for unknown confounders.
Allocation concealment
The person enrolling patients must not know the next allocation (for example sealed opaque envelopes or central telephone or web randomisation). Without it, investigators may consciously or unconsciously steer patients into one group.
Allocation concealment protects the randomisation process (before allocation). Blinding protects against bias after allocation. They are different.
Blinding (masking)
| Type | Who does not know the allocation |
|---|---|
| Single-blind | The participant |
| Double-blind | The participant and the investigator (those giving treatment and assessing outcomes) |
| Triple-blind | Participant, investigator and the person analysing the data |
Blinding prevents performance bias (different care) and detection bias (biased assessment), and reduces the placebo effect influencing reported outcomes.
A placebo is an inactive treatment that looks like the real one. Its use is ethical only when no proven effective treatment exists or when withholding it causes no serious harm.
The Hawthorne effect: people change their behaviour because they know they are being studied.
Analysis
- Intention-to-treat (ITT): everyone is analysed in the group to which they were randomised, whether or not they took the treatment or completed the study. It preserves the benefits of randomisation and reflects real-life effectiveness. Preferred.
- Per-protocol: only those who completed treatment as planned are analysed. May exaggerate benefit, because dropouts are often different.
Intention-to-treat keeps the groups as they were randomised and avoids bias from dropouts.
Results are expressed as relative risk, absolute risk reduction, relative risk reduction and number needed to treat (see measures of risk).
Special trial designs
| Design | Description | Example |
|---|---|---|
| Parallel | Each group gets one treatment throughout | Most RCTs |
| Crossover | Each participant receives both treatments in turn, with a washout period between; each acts as their own control | Stable chronic conditions (for example drugs for chronic pain or hypertension) |
| Factorial | Tests two or more interventions at once in combinations (A, B, A + B, neither) | Aspirin and beta-carotene in the Physicians' Health Study |
| Cluster randomised | Randomises groups (villages, schools, clinics) rather than individuals | Community health education or vaccination programmes |
| Field trial | Healthy people in the field (community setting) receive a preventive measure | Vaccine trials |
| Community trial | Whole communities receive the intervention | Water fluoridation, salt iodisation |
Phases of clinical drug trials
| Phase | Participants | Purpose |
|---|---|---|
| Preclinical | Laboratory and animals | Toxicity, mechanism |
| Phase I | Small number (about 20–80) of healthy volunteers (patients for cancer drugs) | Safety, dose range, pharmacokinetics |
| Phase II | A few hundred patients | Efficacy and side effects; best dose |
| Phase III | Hundreds to thousands of patients in RCTs | Confirm efficacy against standard treatment or placebo; basis for licensing |
| Phase IV | The general population after marketing | Post-marketing surveillance: rare and long-term side effects, new uses |
Rare adverse effects usually appear only in phase IV (post-marketing surveillance), when large numbers use the drug.
Quasi-experimental studies
When randomisation is not possible, practical or ethical, researchers use quasi-experimental designs. The investigator still applies the intervention, but without random allocation.
Types:
- Non-randomised controlled trial: intervention and control groups chosen by convenience (for example two districts).
- Before-and-after (pre-test, post-test) study: outcomes measured before and after the intervention in the same group, with or without a control group.
- Interrupted time series: many measurements before and after an intervention, showing changes in trend (for example after a new law).
- Historical controls: comparing current patients with past ones.
- Natural experiments: an event outside the researcher's control creates comparison groups.
Strengths: practical for evaluating health programmes and policies in real settings.
Weaknesses: groups may differ, so confounding and selection bias are more likely; secular trends can mimic an effect.
The key feature that separates a true experiment (RCT) from a quasi-experiment is randomisation.
Expect: the purpose of randomisation, allocation concealment vs blinding, single, double and triple blinding, intention-to-treat analysis, crossover, factorial and cluster trials, field vs community trials, the phases of drug trials (phase IV for rare side effects), and what defines a quasi-experiment.
Quick revision
- RCT: random allocation; gold standard for treatment questions.
- Randomisation balances known and unknown confounders.
- Allocation concealment before; blinding after allocation.
- Double-blind: participant and investigator unaware.
- Intention-to-treat: analyse as randomised.
- Phase I safety (volunteers); II efficacy; III large RCTs; IV post-marketing.
- Quasi-experiment: intervention without randomisation.
Written by The House Job from Excel Community Medicine and Park's Textbook of Preventive and Social Medicine (named in the UHS PGET syllabus).
FAQ
What is the purpose of randomisation?
To make the groups comparable for both known and unknown factors, so that differences in outcome can be attributed to the intervention.
What is a double-blind trial?
A trial in which neither the participants nor the investigators know who is receiving which treatment.
What is intention-to-treat analysis?
Analysing all participants in the groups to which they were randomised, regardless of whether they completed the treatment.
What is a quasi-experimental study?
A study in which an intervention is applied but participants are not randomly allocated to groups.