RCTs follow a standard design, which randomly allocates patients to one of two or more treatment groups. The standard design for an RCT is illustrated in this flow diagram; click on each item for more information.
Randomised control trials
Patients
A trial will usually focus on a specific type of patient (e.g. those with stroke) or problem (e.g. stroke patients at risk of DVT or pulmonary embolism).
Screening for eligibility
The protocol will define inclusion and exclusion criteria. For example, it might include only those patients who are immobile. It might exclude those who might be harmed by the treatment being tested (e.g. those with peripheral vascular disease).
Consent
Before a patient can be entered into a research study they, or their next of kin (where a patient has mental incapacity), need to be provided with a patient information leaflet and an explanation of the study and be given the opportunity to ask questions about the study.
The patient, or their next of kin, and the person enrolling them need to complete a consent form which is copied and filed away safely. This process must be fully described in the patient’s medical record.
Participants in research always have the option of withdrawing from the study without having to give a reason.
Baseline data collection
This includes: information about the patient (e.g. age and sex), their medical condition (e.g. diagnosis, severity) and sometimes contact details to allow the researchers to follow the patients up. These data are usually entered onto a paper form and/or a secure web page.
Randomisation
Patients are randomly (like the toss of a coin) allocated to receive one treatment or another. This is usually done by a computer to ensure that neither the patient or the person entering the patients can guess which treatment they will receive before the patient is randomised.
If large numbers of patients are randomised, the two different treatment groups will have very similar characteristics (e.g. very similar average age, stroke severity etc) except for the treatment they receive.
Treatment A
This might be the treatment being tested (e.g. aspirin, long anti-embolism stockings, admission to a stroke unit). Staff need to ensure that patients actually receive the allocated treatment, or at least as much of it as possible.
Follow up
All the patients in each treatment group are followed up for a defined length of time and assessed at pre-defined intervals. The follow up for those in each group should be identical and ideally be performed by a person who is blind to their treatment allocation.
Even if patients do not receive the allocated treatment (for example they develop adverse effects or refuse to take the treatment ) they should be followed up if possible.
Patient outcomes
This is a measure of how the patients have done and will differ depending on the purpose of the treatment being tested. For instance, it might include survival, a measure of independence, whether the patient has had a DVT or a bleed. The patient outcomes should ideally be assessed by someone who does not know which treatment the patient received (blinded) and using a reliable measure (e.g. a questionnaire or scan).
In a randomised trial any “significant” difference in the outcomes between the treatment groups should reflect the relative effectiveness of those treatment. Whether the difference is significant depends on its size and also the number of patients in the study. A statistician will calculate whether any difference seen is likely to reflect a real difference in treatment effectiveness or if it might be due to chance.
Treatment B
This might simply be usual care (that is care without treatment A – e.g. no stockings), or it may be an alternative (e.g. short stockings) or a placebo (e.g. a dummy tablet which looks like aspirin but contains no aspirin).
A placebo treatment means that neither the patient or the staff caring for them are aware of which treatment the patient is receiving. They are described as being “blinded” to the treatment. Blinding reduces the risk of bias because knowledge of the treatment patients are actually receiving can influence their other care and assessments. Sometimes it is difficult to blind patients/therapists in certain trials, for example, a trial testing graduated compression stockings.
Page last reviewed: 14 May 2020
