21 Flashcards
Three essential elements to a RCT
Randomised:
Participants randomly allocated to groups
Controlled:
There is a control (comparison) group
Trial:
Testing effect of treatments/interventions
Randomisation
Cohort v RCT
Cohort
Ascertain exposure status, then follow-up to find out outcome(s)
Observational study
RCT
Randomly assign exposure, then follow-up to find out outcome(s)
Intervention (experimental) study
What is Randomisation
Participants are randomly allocated to groups
Why randomise?
“ Random allocation (also called randomisation) of individuals in the study groups is the only way to ensure that all of the groups are as similar as possible (i.e. exchangeable) at the start of the study”
Randomisation / Random Allocation
- how does it prevent confounding
- is enough people randomly allocated, should have the same proportion of the confounded in each group
- this applied to known and unknown con founders
- successful randomisation means confounding is an unlikely reason for differences in outcomes between groups
How do we know if randomisation has been successful?
Randomisation / random allocation ≠ random selection
Variants e.g. cluster randomisation
- RCT
It may be difficult to randomise individuals – randomise groups (clusters) of participants instead e.g. GP practices, hospital wards, schools
Preserving the benefits of randomisation - different types of analysis that allows this / dont allow this
Intention-to-treat analysis
- analyse as randomised
- preserves the benefits of randomisation (- regardless whether or not they stuck to that group)
Pre-protocol analysis
- analyse as treated (- treated according to what they actually did )
- tends to be more appropriate for efficacy trials (does this medication work if u take it)
- introduces confounding
Potential sources of bias in RCTs
• Lack of concealment of allocation
• Lack of blinding
• Loss to follow-up
• Non-adherence
Concealment of allocation
- what does this mean in preventing bias in a RCT
Important that allocation sequence is concealed and unpredictable
Otherwise could game the system and introduce bias
Blinding in RCT
How might knowing which group/arm of the study a participant was in affect:
A research team member
- delivering the intervention?
- assessing the outcome?
- analysing the data?
The participant?
State who is blinded (rather than ‘single’ or ‘double’ blinded)
What is blinding ?
Blinding is an important way to protect against bias
Why blinding might be challenging to active in practice?
But can be challenging to achieve in practice…
• It can be obvious which group participant is in:
e.g. you might be able to get a matching placebo pill but what about surgical or physiotherapy treatment?
• Safety concerns
Non-adherence
Participants don’t always do what you ask them to do
Strength of RCTs
Randomisation (if done well) maximises the likelihood that potential confounders are evenly distributed between groups
Best way to test an intervention
Can calculate incidence ( so can calculate relative risk and risk difference)
Temporal sequence is known
Strongest design for determining causation
Limitations of RCTS
• Not all exposures are suitable to be randomly allocated
• Need to have clinical equipoise
Not all exposures are suitable to be randomly allocated
- why
Some exposures cannot be randomly allocated:
• Low birth weight
Some exposures are very difficult to randomly allocate:
• Long-term behavioural changes
• Common exposures in the community (how do you avoid them?)
Some exposures should not be randomly allocated:
• Known harmful toxins or procedures
What is clinical equipoise
Genuine uncertainty about benefit or harm of the intervention
Why do we need to have clinical equipoise
Unethical to:
• Give a known harmful intervention to people
• Give interventions known to be less effective than current
treatments
• Waste resources and risk people’s well-being if you already
know the answer
Limitations of RCT
• Can be difficult to achieve blinding
• Potential for loss-to-follow up and non-adherence
• Resource intensive
Highly selective
Often a highly selected group of people involved in an RCT
• They need to meet all the inclusion criteria
• AND be willing to participate
This can affect generalisability (sometimes called external validity) – do the findings apply to other people?
