Randomised controlled trials

Wikis > Research > Types of studies > Experimental Studies > Randomised controlled trials

Randomised controlled trials (clinical trials):
• An experiment in which subjects in a population are randomly allocated into groups – the ‘treatment’ and ‘control’ group – to receive an intervention or a ‘non-intervention’ – the outcomes in the two groups are compared
• very ‘powerful’ research design
• To ensure equivalence between the groups, subjects are randomly allocated to the groups – within the limits of chance this ensures that all groups will be comparable at the start of the study – it is assumed that randomisation will make the groups equivalent in known and unknown variables, so outcomes measure the effect of the intervention. Random numbers can be obtained from published statistical table or generated by computer.
• The form of intervention whose effects are being tested is called the independent variable, while the outcome being measured is the dependent variable or response variable
• Assessments of subjects should be done by those who are blind to which group the subject is in. Similarly, the subject should be blind to which group they are in or if they are in the intervention or placebo group.
• Types Blinding-
a) Subject binding:
- Subject unaware of which group they are in (which intervention they have)
b) Observer blinding:
- Procedures put in place so observers/assessors can be as objective as is possible when measuring outcomes
• Intention to Treat – if subjects withdraw and do not complete treatment  problems for analysis as composition of control group may no longer be free from bias  analyse data assuming each subject got and completed treatment the way initially intended  this make it harder to find a statistical difference. Additional analyses may be subsequently performed on those patients who did complete the trial.
• Pragmatic Trial- adaptation of RCT to assess intervention under conditions similar to clinical practice. Blinding but not always possible. Often used to determine the best t/t for a particular group.

Comments are closed.