Randomised Control Trials Flashcards
(8 cards)
Why do we do RCTs?
Well designed trials are necessary to gather robust evidence as to what works and what doesn’t, what makes a difference and what doesn’t. They establish the average efficacy of a treatment across a population. Designed to control for change not due to treatment e.g. spontaneous recovery, maturation.
An RCT should eliminate most of the key biases which can challenge the validity of intervention research. Study quality still matters - can be poorly conducted or well conducted like all other types of research, BUT they have many advantages.
What are the key features of RCTs?
Study participants randomly allocated to treatment groups. Neither research staff or participants influence who gets what treatment- reducing bias by equalising other factors not accounted for in the design study e.g. severity, SES.
The population of sample clearly defined
Results only generalise to same population
Treatments clearly defined. Others should be able to use the same treatment if it is effective.
The outcome measures should be clearly defined and relevant to the kinds of benefits anticipated.
They should show what kind of benefit and the degree of benefit that can be expected from the intervention
The biggest advantage of RCTs?
Reducing biases of intervention studies.
How does RCT reduce bias?
Individuals/ clusters of individuals allocated randomly to intervention/control/treatment groups.
This results in “parity” or equivalence at baseline/ equivalence in groups. All other things being equal, it is the intervention which is causing the change in the outcome.
Often referred to as a truly experimental design because the variable that you are manipulating is what causes the change (everything else remains equal).
Often contrasted with quasi-experimental or observational designs (case-studies) both of which are thought to be prone to bias.
Additional advantages of RCTS
RCTs are designed to establish the average efficacy of a treatment across a large population and learn about frequently occurring side-effects. Treatment effects may vary between people, making inference from single case studies unreliable.
Treatment effects can be small and therefore undetectable except when studied systematically on a large sample.
RCTs are designed to control for change which might have been due to spontaneous recovery unrelated to treatment; maturation, in the case of children; other unexpected things.
Drawbacks of RCT
RCTs are more costly and tome consuming than other studies.
RCTs often have highly restricted selection criteria for participants.
You need to estimate sample size and justify expense of the trial. An estimated large treatment effect is often warranted in order to justify the time an expense of an RCT.
The intervention needs to be thoroughly developed to be evaluated effectively- often preliminary research (pilot study) is needed to justify the choice of intervention.
Results can only be generalised to the people that the sample population relates to. Cannot pick up RCT on men over 60 who have responded well to aphasia treatment and generalise to young adults with aphasia.
People with communication disorders are typically heterogenous; therapy methods are heterogenous. If you must exclude many to participate in an RCT then you may be losing a large proportion of the target population. RCTs are therefore difficult in SLT.
Some RCTs have shown some people made no improvement at all, where others made lots of progress – but this results in the study reporting that on average there is an improvement. But as a clinical SLT you would be worried about the group that made no improvement, who are they? Can we take these results to inform all our practice?
Evaluating RCTs
Just because an RCT has been done doesn’t guarantee that it was done well. There are certain rules to follow. The CONSORT checklist can be used when reporting an RCT and when critically appraising one. You can also use the PICO method to focus on certain aspects of the study.
RCTs should be blinded either single or double- what does this mean?
Single blinded studies are where the researcher knows the details of the treatment, but the patient does not. This eliminates placebo effects, but observer bias is possible.
Double-blind studies are where the clinician, assessor and patient do not know which group the patient was assigned to. Not always possible or practical, especially with behavioural interventions.
Blind assessors are very important however and should be incorporated.
